home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
CD ROM Paradise Collection 4
/
CD ROM Paradise Collection 4 1995 Nov.iso
/
science
/
simst35.zip
/
INSTALL._D2
/
DEMO.SCR
< prev
next >
Wrap
Text File
|
1994-11-05
|
46KB
|
1,797 lines
:START;
OPEN SAMPLE.SPS;
CLS;
*SET DEMO (2000); { to run in demo mode with a 2 second delay }
SET COLOR BLACK ON CYAN;
IF $MONO SET COLOR BLACK ON LIGHTGRAY;
:MAINMENU;
ON ESC GOTO MAINMENU;
CLS;
IBOX ROW 4 COLOR BLACK ON LIGHTGRAY
Welcome to the SIMSTAT demonstration program.
Select one of the following commands then press the ENTER key.
ENDBOX;
MENU ROW 11
&Overview of SIMSTAT
&Bootstrap analysis
&New features of v3.5
&Command language
b&Anner v1.0
&Registering SIMSTAT
&Quit
ENDMENU;
IF $ITEM = 1 GOTO DEMO;
IF $ITEM = 2 GOTO BOOT;
IF $ITEM = 3 GOTO NEW;
IF $ITEM = 4 GOTO LANGUAGE;
IF $ITEM = 5 GOTO BANNER;
IF $ITEM = 6 GOTO ORDER;
IF $ITEM = 7 STOP;
GOTO MAINMENU;
**************** DEMO *****************;
:DEMO;
OPEN SAMPLE.SPS;
CLS;
BOX
SIMSTAT v3.5 is an easy to use statistical program
that provides a wide range of statistical analyses,
high-resolution graphs, and powerful simulation tools.
ENDBOX;
BOX
One of the most useful features of SIMSTAT is its ability
to perform analyses on various data file formats such as:
■ dbase II, II or IV (.DBF)
■ Lotus 1-2-3 or Symphony (.WKS or WK1)
■ Comma-separated value ASCII (.DAT or CSV)
■ Fixed format ASCII (.FIX)
■ SPSS/PC+ (.SPS, .SYS or SIM)
■ SPSS for Windows (.SAV)
ENDBOX;
BOX
For the next few minutes, we will take a tour that will show you
the various features of SIMSTAT. To interrupt the demonstration
simply press the Escape key.
ENDBOX;
LIST SEX TO AGGRESS /N 15;
BOX ROW 11
Once you have entered your data in a file, the first thing you
will want to do is to check the values in the file to identify
illegal values. The LIST command displays the values of each
variable previously selected. This command lists the content
of all cases or can be restricted to a specific number of cases.
ENDBOX;
DESC SEX TO AGGRESS;
BOX ROW 9
The DESCRIPTIVE command allows to quickly display
summary statistics on all your variables.
ENDBOX;
BOX ROW 8
Besides the mean and the standard deviation which give
a quick overview of the data distribution, the minimum
and maximum values can also be used to identify values
out of range while the number of valid cases allows you
to see how many cases are missing.
ENDBOX;
BOX
More detailed statistics can be obtained by using the
FREQUENCY command. Let us look at them one at the time.
ENDBOX;
FREQUENCY AGE
/TABLE;
BOX ROW 3
The FREQUENCY command allows you to display a table of frequencies that
shows the frequency and the percentage of each value in a variable.
ENDBOX;
BOX COL 1 ROW 11
While this table is
sorted in ascending
order of values, it is
also possible to sort
it in descending order
of frequency.
ENDBOX;
FREQUENCY AGE /DESC CI=80;
BOX ROW=5
The FREQUENCY command also displays detailed descriptive
statistics, including measures of central tendency and
of dispersion. Other statistics are useful to get a
glimpse of the distribution shape.
ENDBOX;
FREQUENCY AGE /PTILES=10;
BOX ROW 10
The percentiles option provides a more detailed description
of the variable's distribution. You can specify the number
of categories you want from 2 to 100 categories.
ENDBOX;
FREQUENCY AGE /PTILES=4;
BOX ROW 12
By setting the number of categories to 4, the
percentiles option produces the interquartile
values also know as Q1, Q2 (or median) and Q3.
ENDBOX;
BOX
As you may have noticed, sometimes the results
of a statistical analysis scroll off the screen.
SIMSTAT allows you to browse back through the
previous analyses by using the F2 key. Let us
browse through what we have done so far.
ENDBOX;
BOX
Use the navigation keys to move around the listing.
When you have finished, press the Escape key.
ENDBOX;
BROWSE;
BOX
A graphic representation is also an efficient
way to examine your data and a powerful means
to communicate them to others.
The FREQUENCY command allows you to obtain
various descriptive graphs such as...
ENDBOX;
QBOX "a bar chart";
FREQUENCY AGE /BARCHART=G;
QBOX "a pie chart" ;
FREQUENCY AGE /PIE;
QBOX "a Pareto chart";
FREQUENCY AGE /PARETO;
BOX;
The preceding graphs were appropriate to display
information on categorical variables. But when
the data are quantitative we often use other
types of graphs such as...
ENDBOX;
QBOX "an histogram,";
FREQUENCY AGGRESS /HISTO NORMAL NBAR=20;
QBOX "a box-&-whisker plot,";
FREQUENCY AGGRESS /BOXPLOT;
QBOX "a cumulative distribution chart,";
FREQUENCY AGGRESS /CUMUL;
QBOX "or a normal probability plot.";
FREQUENCY AGGRESS /PPLOT;
CLS;
BOX
Each component of high resolution graphs such as the
title, the axis labels, and scales can be edited. You
can also add a 3D effect on bars, display horizontal
or vertical grids, and change the color, size, and
appearance of almost any graph component.
ENDBOX;
BOX
Graphs can also be printed on almost any
dot matrix, inkjet or laser printer,
including postcript and color printers.
ENDBOX;
BOX
They can also be saved on disk in PCX or HPGL
format, to be imported in your favorite word
processor or illustration program.
ENDBOX;
CROSSTAB SIBLING BY SEX
/TABLE DVALUE;
BOX COL=51
The CROSSTAB command
is useful to identify
relationships between
categorical and/or
ordinal variables.
ENDBOX;
BOX COL= 51
A standard table produced
by SIMSTAT shows the
number of cases in each
cell of the table.
ENDBOX;
BOX
But you can also ask SIMSTAT to include
more detailed information such as:
■ row percentages
■ column percentages
■ total percentages
■ expected values
■ chi-square residuals
■ standardized chi-square residuals
ENDBOX;
QBOX "Let us see a table with row and column percentages";
CROSSTAB SIBLING BY SEX
/TABLE CELLS=12;
QBOX "Here it is!" COL 55;
BOX
To test whether there is a relationship between
the two variables or to assess the strength of
this relationship, a wide range of statistics can
be obtained. Some of those statistics are useful
when both variables are nominal, while others are
more appropriate when both variables are ordinal
or continuous.
Let us look at all those statistics!
ENDBOX;
CROSSTAB SIBLING BY SEX
/TABLE /STAT=123;
QBOX "Here are some statistics used when both variables are nominal" ROW 3;
CROSSTAB SIBLING BY SEX
/TABLE /STAT=456789;
QBOX "While the following are used when the data are ordinal" ROW 6;
BOX
To visualize the relationship between two categorical
variables SIMSTAT can display a bar chart.
ENDBOX;
GRAPH 3D OFF OVERLAP 0;
CROSSTAB AGE BY SEX
/BARCHART;
CLS;
BOX
An option panel lets you control the final appearance
of the graph. For instance, here is the same graph
but with a 3D effect and a 50% bar overlap.
ENDBOX;
GRAPH 3D ON OVERLAP 50;
CROSSTAB AGE BY SEX
/BARCHART;
BOX
The MULTIPLE RESPONSE procedure allows you to obtain
frequency analyses and crosstabulation analyses on
variables which can legitimately have more than one
response. These multiple responses can be stored in
as many variables as necessary. SIMSTAT will gather
all these responses and treat them as if they were
stored in a single variable.
ENDBOX;
MRESPONSE MULTY
/FREQUENCY SEX SIBLING /TABLE;
BOX ROW 2
Beside existing options, graphs and analysis usually available
in the FREQUENCY and CROSSTAB command, the MULTIPLE RESPONSE
computes percentages on the total number of cases and on the
number of valid responses.
ENDBOX;
CLS;
BREAKDOWN AGGRESS BY SEX
/RANGE (1 2);
BOX ROW 16
The BREAKDOWN command can be used to display the value
of a quantitative variable among various groups defined
by a categorical variable. The default format outputs
a single line of statistics for each group.
ENDBOX;
BREAKDOWN AGGRESS BY SEX /RANGE (1 2) DETAIL;
BOX ROW 4
However, you may also obtain more detailed
statistics for each subgroup of subjects.
ENDBOX;
BOX
The BREAKDOWN command also allows you to
examine the distribution of each group by
displaying a multiple box-&-whisker plot.
ENDBOX;
BREAKDOWN AGGRESS BY SEX /RANGE (1 2) BOXPLOT=G;
BOX
While the BREAKDOWN command is useful to identify
differences between subgroups on a quantitative
variable, it cannot be used to decide whether those
data suggest a real difference in the population.
ENDBOX;
BOX
SIMSTAT provides numerous analyses appropriate for
answering this question. The most common are the
T-TEST and the one-way analysis of variance (or ANOVA).
ENDBOX;
BOX
SIMSTAT provides two types of t-test analysis.
ENDBOX
T-TEST AGGRESS BY SEX
/GROUP VALUE 1,2 NOEFFECT;
BOX ROW 10 COL 55
An independent or
grouped t-test for
comparing the score
on a quantitative
variable between
two groups defined
by a dichotomous
variable.
ENDBOX;
T-TEST HOURSTV BY AGGRESS /PAIRED NOEFFECT;
BOX ROW 15 COL 52
And a paired or
correlated t-test
to compare the means
of a single group on
two variables or
of two groups of
paired subjects.
ENDBOX;
T-TEST AGGRESS BY SEX
/GROUP VALUE 1,2;
BOX ROW 6
Beside descriptive statistics, homogeneity of variance
test, and standard t-test statistics, SIMSTAT also displays
various effect size measures allowing you to assess the
size of the difference between the two groups.
ENDBOX;
BOX
When performing a t-test, three kinds of graph
can be obtained to visualize the difference:
ENDBOX;
BOX
a DUAL HISTOGRAM to inspect in detail the data
distribution in both groups. This graph can be
displayed horizontally,
ENDBOX;
T-TEST AGGRESS BY SEX
/GROUP VALUE (1 2) HISTO NORMAL;
CLS;
QBOX "or vertically";
T-TEST AGGRESS BY SEX
/GROUP VALUE (1 2) HISTO VERTICAL;
CLS;
BOX
a bar chart representing the mean of both groups with
an option to display an error bar representing either
the standard deviation, the standard error or a user
defined confidence interval.
ENDBOX;
T-TEST AGGRESS BY SEX
/GROUP VALUE (1 2) BARCHART CIBAR CI=90;
CLS;
BOX
an finally, an error bar diagram which provides
another way to display the same information.
ENDBOX;
T-TEST AGGRESS BY SEX /VALUE (1 2) ERRORCHART CI=90;
CLS;
ONEWAY AGGRESS BY SIBLING;
BOX
The ONEWAY anova command allows you to compare
the means of a quantitative variable for two
or more independent groups.
ENDBOX;
BOX ROW 4
Beside the standard ANOVA table, SIMSTAT
provides much useful information such as...
ENDBOX;
BOX ROW 14
the proportion of explained variance (or R square)
which shows the strength of the relationship,
ENDBOX;
BOX ROW 17
a test of homogeneity of variance,
ENDBOX;
ONEWAY AGGRESS BY SIBLING /DESC SCHEFFE;
BOX ROW 3
descriptive statistics for each group,
ENDBOX;
BOX ROW 6
And various multiple comparison procedures such as:
■ Least Significant Difference (LSD)
■ Scheffé
■ Tukey's HSD (Honestly Significant Difference)
■ Newman-Keuls
ENDBOX;
BOX
The ONEWAY command also allows you to obtain graphical
presentation of the data for each group such as...
ENDBOX;
BOX
a barchart representing the mean of each group with
an option to display an error bar representing
either the standard deviation, the standard error
or a user defined confidence interval.
ENDBOX;
ONEWAY AGGRESS BY AGE /BARCHART SD;
CLS;
QBOX "an error bar diagram,";
ONEWAY AGGRESS BY AGE /ERRORCHART SD;
CLS;
BOX
and a deviation barchart which expresses
the deviation of the mean of each group
from the grand mean.
ENDBOX;
ONEWAY AGGRESS BY AGE /DEVCHART;
CLS;
BOX
SIMSTAT also provides various tools to examine
the relationship between two or more quantitative
variables such as correlation matrix, linear,
nonlinear, and multiple regression analysis.
ENDBOX;
CORRELATION AGE AGGRESS HOURSTV /EXACT 2TAIL;
BOX COL 48
The CORRELATION command
provides a quick look at
the strength of the linear
relationship between a
number of quantitative
variables.
ENDBOX;
{numerous quantitative variable vs a number of}
CORRELATION AGE AGGRESS HOURSTV /2TAIL COVAR;
BOX ROW 4
Beside correlation coefficients and their associated
probabilities, SIMSTAT can also display a cross-product
and variance-covariance matrix.
ENDBOX;
CORRELATION AGE AGGRESS HOURSTV /2TAIL CI=90;
BOX COL 48
User-specified confidence
intervals can also be
included in the matrix.
ENDBOX;
BOX
SIMSTAT also allows you to graphically examine
the relationships between numerous variables
with the use of a scatterplot matrix.
ENDBOX;
CORRELATION NO AGE AGGRESS HOURSTV /XYPLOT TREND;
REGRESSION AGGRESS BY HOURSTV /LINEAR 2TAIL CI=80;
BOX
The REGRESSION analysis can be used to obtain
more detailed information about the relationship
between two quantitative variables.
ENDBOX;
BOX ROW 3
The information displayed includes the correlation
coefficient and the proportion of variance explained.
ENDBOX;
BOX ROW 7
an anova table,
ENDBOX;
BOX ROW 11
and various information about the regression equation
such as the values of the parameters, their confidence
intervals, their probability, etc..
ENDBOX;
BOX
This procedure can also display a text
and/or graphic scatterplot of the
relationship between the variables;
ENDBOX;
REGRESSION AGGRESS BY HOURSTV /LINEAR 2TAIL CI=80 XYPLOT=G;
BOX
Beside linear regression, the REGRESSION command
also performs basic nonlinear regression analysis
such as quadratic, cubic, 4th and 5th degree
polynomials, exponential and inverse relationship.
ENDBOX;
BOX
For instance, here is an example of a quadratic
relationship between two quantitative variables.
ENDBOX;
REGRESSION AGGRESS BY AGE
/QUADRATIC 2TAIL CI=80 XYPLOT=G;
REGRESSION AGGRESS BY HOURSTV /CPLOT OUTLIERS 1.5;
BOX ROW 3
SIMSTAT also provides many tools to examine the
residuals and evaluate the appropriateness of the
model such as a caseplot of standardized residuals,
ENDBOX
BOX
a scatterplot of predicted values
against standardized residual,
ENDBOX;
REGRESSION AGGRESS BY HOURSTV /RPLOT=G;
BOX
and a normal probability plot
of the residual values.
ENDBOX;
REGRESSION AGGRESS BY HOURSTV /PPLOT;
BOX
Predicted and standardized residual values can also
be saved in a separate file for further analysis.
ENDBOX;
BOX
SIMSTAT also includes a powerful
multiple regression procedure in
which the user can choose among
various regression methods such as:
■ standard regression
■ forward selection
■ backward elimination
■ stepwise selection
■ hierarchical entry.
ENDBOX;
MULTREG AGGRESS BY AGE SEX HOURSTV
/HIERARCHICAL
/ORDER (AGE SEX) (HOURSTV)
/EQUATION OUT ANOVA CHANGE SUMMARY CI=80;
BOX
In addition to all the statistics available in
the REGRESSION command, it is also possible obtain
detailed statistics at each step of the analysis
such as a statistical test of R changes, the
tolerance of variables in the model, the
probability of variables not entered, etc..
ENDBOX;
MULTREG AGGRESS BY AGE SEX HOURSTV
/STEPWISE PIN (0.05) POUT (0.055) TOLERANCE (0.001)
/ANOVA EQUATION HISTORY
/RPLOT=G
/PPLOT;
CLS;
BOX
The GLM ANOVA/ANCOVA procedure is a General Linear Models
implementation of analysis of variance and covariance. It
can handle balanced and unbalanced ANOVA designs and support
models with categorical and/or quantitative variables. The
procedure can also be used to perform standard multiple
regression problems that involve interaction terms.
ENDBOX;
BOX
The options panel allows you to display
standard ANOVA/ANCOVA tables as well as
various outputs usually found in ANOVA/ANCOVA
or multiple regression analyses.
ENDBOX;
SET MORE ON;
GLMANOVA AGGRESS BY AGE SEX SIBLING
/NONEXPERIMENTAL
/QUANT AGE
/INTERACTION AGE*SEX
/STEP MEAN CHANGE RPLOT=G;
SET MORE OFF;
BOX
SIMSTAT provides various methods of adjustment
for unequal cell size including a hierarchical
strategy where you can set the order of entry
of each variable in the model.
ENDBOX;
SET MORE ON;
GLMANOVA AGGRESS BY SEX SIBLING AGE HOURSTV
/HIERARCHICAL
/QUANT AGE HOURSTV
/ORDER (SEX SIBLING AGE) (HOURSTV)
/MULTREG STEP CHANGE EQUATION CI (95)
/CPLOT OUTLIERS (1.5) CI (95) PPLOT;
SET MORE OFF;
CLS;
BOX
The TIME SERIES command allows the examination
of time series. The options panel offers various
transformations to remove trends or seasonal
dependence in a series and provides diagnostics
for those transformations by displaying
autocorrelation and partial autocorrelation
function plots of the transformed series.
ENDBOX;
BOX
This panel also allows the application of two smoothing methods
(moving average and running median) to identify trends in noisy
time series data. Control bars representing the mean and the
confidence limits can also be displayed over the series.
ENDBOX;
SET MORE ON;
TIME-SERIES AGGRESS
/MEAN
/ACF PACF PLOT
/MAVG (3 4 5 3)
/RBAR PCT (95);
SET MORE OFF;
CLS;
BOX
The SINGLE CASE command provides some basic tools to study
the effect of an intervention on the behavior of a single
subject. It involves the repeated objective measurement
of the behavior of a single subject (dependent variable)
over a long period of time interspersed with changes in
the treatment condition (independent variable).
ENDBOX;
BOX
The procedure will display a graph representing the
evolution of the dependent variable (Y) at various
phases defined by the independent (X) variable. The
options panel allows you to obtain various statistics
for each phase of the analysis as well as various
graphic tools that can be used as judgement aids for
identifying the experimental effect of the intervention
(smoothed data, split-middle trend, control bars).
ENDBOX;
SCED BEHAVIOR BY PHASE
/DETAIL
/TREND;
BOX ROW 5
Various statistics can also be obtained
on each treatment condition.
ENDBOX;
CLS;
BOX
The RELIABILITY command provides a means to assess the
quality of multiple-item additive scales through the
computation of reliability statistics.
ENDBOX;
BOX
The options panel offers the possibility to produce
various item statistics (e.g.: mean, minimum, maximum,
standard deviation), iter-item variance-covariance and
correlation matrices, total scale and item-total statistics.
It also allows you to verify the reliability of the scale
through the use of a split-half method or by computing
internal consistency measures.
ENDBOX;
BOX
Each selected variable is considered as a single item of
the scale. The Xs and Ys are used in the split-half method
to specify how the various items should be divided.
ENDBOX;
DATA
DEPRESS1,DEPRESS2,DEPRESS3,DEPRESS4,DEPRESS5,DEPRESS6,DEPRESS7,DEPRESS8
1 2 1 1 1 2 1 2
2 1 1 2 1 1 1 2
1 1 1 1 1 2 1 1
1 1 1 2 1 1 1 1
1 1 1 1 1 2 1 2
1 2 1 1 1 2 2 2
1 2 2 1 1 2 1 2
1 1 2 2 1 1 1 2
1 2 1 1 1 1 1 2
1 1 1 1 1 1 1 1
1 2 2 1 1 2 3 2
1 1 1 1 1 1 1 2
1 3 1 1 1 1 2 1
1 1 1 1 2 1 1 1
2 2 2 1 2 1 2 2
1 2 2 1 2 2 1 2
1 1 1 1 1 2 1 1
1 2 1 1 1 2 2 2
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
2 2 1 1 1 1 1 1
1 1 1 1 1 2 1 2
1 2 1 2 1 1 2 1
1 2 1 1 1 1 1 2
1 1 1 1 1 2 1 1
1 1 1 1 2 1 1 1
2 2 1 1 2 2 1 2
1 2 1 1 1 1 1 1
1 1 1 1 1 1 1 2
1 1 1 1 1 2 1 1
1 1 1 1 1 2 1 2
1 2 1 1 2 2 1 2
1 2 1 1 1 1 2 2
1 1 1 1 1 1 1 2
1 2 2 1 1 3 3 2
1 2 1 1 1 1 1 1
1 2 2 1 1 2 1 2
1 2 2 1 1 2 1 3
1 1 1 1 1 2 1 1
1 2 1 1 1 2 1 1
1 1 1 1 1 1 1 2
1 2 2 1 2 3 2 2
2 2 1 2 1 2 2 3
1 2 1 1 1 1 1 2
2 2 2 2 2 2 1 2
1 1 1 1 1 2 1 1
2 2 2 2 1 2 2 2
1 2 1 1 1 2 1 2
1 2 2 1 1 2 1 2
1 2 1 1 1 2 1 2
ENDDATA;
SET MORE ON;
RELIABILITY DEPRESS1 TO DEPRESS4 BY DEPRESS5 TO DEPRESS8
/ITEM CORR COVAR TOTAL ALPHA SPLIT;
PAUSE;
OPEN SAMPLE.SPS;
CLS;
BOX
The SENSITIVITY analysis allows one to assess
the ability of a quantitative measure (X) to
differentiate a dichotomous criterion condition
(Y) and provide guidelines to choose an
appropriate cutoff point.
ENDBOX;
SENSITIVITY SEX BY AGGRESS
/VALUE (1) HIGH SSTAT ESTAT;
SET MORE OFF;
BOX
The program provides for each value of the quantitative
measure the level of sensitivity (proportion of positive
cases correctly diagnosed as true) and specificity
(proportion of negative cases correctly diagnosed as false),
and the percentage of false-positives and false-negatives.
ENDBOX;
BOX
The options panel also allows you to obtain a receiver-
operating-characteristic (ROC) curve and Error rate graph.
ENDBOX;
SENSITIVITY SEX BY AGGRESS
/VALUE (1) HIGH ROC ERROR;
CLS;
BOX
SIMSTAT also performs a wide range
of nonparametric tests. Lets take
a look at those test.
ENDBOX;
BINOMIAL AGE
/VALUE 10 PROPORTION .50;
BOX ROW 18
The BINOMIAL TEST allows you to assess whether
the observed number of cases in a dichotomous
variable is the same as that expected from a
specified binomial distribution
ENDBOX;
CHISQUARE SEX
/VALUE (1 2)
/FREQ (40 60);
BOX ROW 6
The ONE SAMPLE CHI-SQUARE TEST allows you to assess
whether there is a difference between the observed
number of cases in various categories and the
expected frequencies in those same categories.
ENDBOX;
RUNSTEST HOURSTV /MEAN;
BOX COL 42
The RUNS TEST to test whether
the ordered sequence in which
observations were obtained
is random.
ENDBOX;
MCNEMAR SEX BY SIBLING
/VALUE (1,2) 2TAIL;
BOX ROW 3
The McNEMAR TEST is a procedure applied to a pair of
correlated dichotomous variables to test whether there is a
significant difference in proportions of subjects that
change from one category to another.
ENDBOX;
MANN-WHITNEY AGGRESS BY SEX /VALUE (1 2) 2TAIL;
BOX COL 44
The MANN-WHITNEY U test
procedure evaluates the
hypothesis that two
independent samples have
the same distribution.
The Mann-Whitney U is the
nonparametric version of
the t-test for independent
samples.
ENDBOX;
WILCOXON AGGRESS BY HOURSTV
/2TAIL;
BOX ROW 2
The WILCOXON matched-pairs signed-ranks test is a procedure used
to test whether two related samples have been drawn from the same
population. It takes into account the magnitude as well as the
direction of the differences between the two variables.
ENDBOX;
SIGN-TEST AGGRESS BY HOURSTV
/2TAIL;
BOX ROW 4
SIGN TEST tests the hypothesis that two variables
have the same distribution. This is assessed by
comparison of the numbers of positive and negative
differences between the two variables.
ENDBOX;
KS1 AGGRESS /NORMAL;
BOX ROW 8
The KOLMOGOROV-SMIRNOV ONE-SAMPLE TEST compares the
distribution of each variable against a standard normal
distribution or a uniform distribution. It tests whether
the sample data can reasonably be thought to have come from
a population having this theoretical distribution.
ENDBOX;
KS2 AGGRESS BY SEX
/VALUE (1 2);
BOX COL 45
The KOLMOGOROV-SMIRNOV
TWO-SAMPLE TEST evaluates
whether a variable (Y) has
the same distribution in two
independent samples as defined
by a grouping variable (X).
This test is sensitive to
differences in the shape,
location, and scale of the
two sample distributions.
ENDBOX;
MOSES AGGRESS BY SEX
/VALUE (1 2) OUTLIERS (5);
BOX COL 42
The MOSES TEST of extreme
reactions tests whether the
range of an ordinal variable
is the same in a control
group as in a comparison
group, as defined by a
grouping variable.
ENDBOX;
KRUSKAL AGGRESS BY AGE
/RANGE (6 11);
BOX COL 42
The KRUSKAL-WALLIS one-way
analysis of variance by ranks
is a procedure for testing
whether k groups have been
drawn from the same population.
This test is a nonparametric
version of the one-way analysis
of variance.
ENDBOX;
MEDIAN AGGRESS BY SEX /VALUE (1 2);
BOX ROW = 4
The MEDIAN TEST is a procedure for testing whether two or
more independent groups differ in central tendencies. It
tests the likelihood that those groups were drawn from
populations with the same median.
ENDBOX;
*MEDIAN AGGRESS BY AGE /EXTENDED VALUE (6 11);
*QBOX "Median test for k samples...";
FRIEDMAN AGGRESS HOURSTV AGE;
BOX ROW 7
The FRIEDMAN TEST is a procedure for testing
wether two or more related samples have been
drawn from the same population.
ENDBOX;
NPAR AGGRESS HOURSTV AGE NO
/TAU-B EXACT 2TAIL LISTWISE;
BOX
The NPAR MATRIX displays a matrix for various measures
of association and concordance between two variables
(i.e. Spearman's Rs, Kendall's Tau-a and Tau-b, Kendall-
Stuart's Tau-c, symmetric and asymmetric Somer's D and
Goodman-Kruskal's Gamma).
ENDBOX;
CLS;
BOX
One of the most innovative features of SIMSTAT is
the inclusion of powerful bootstrap resampling
analysis. To learn more about this revolutionary
new method, select the Bootstrap analysis from the
main menu.
ENDBOX;
BOX
Simstat also provides a lot of features useful for statistical
analysis such as:
■ conditional selection of cases
■ weighting of cases
■ user defined missing values on all type of files
■ variable and value labels
ENDBOX;
BOX
While SIMSTAT can be configured to automatically
call your own data editor, it also provides a
powerful spreadsheet for its own file format and
for SPSS files.
ENDBOX;
BOX
This editor named SIMEDIT allows you to import and export
most data file formats supported by SIMSTAT, modify the
values, perform conditional transformation, ranking and
recoding of data, compute new variables, and add missing
values, variable and value labels.
ENDBOX;
BOX
SIMEDIT includes 47 functions and operations for data
transformation including mathematic, trigonometric functions,
statistical operations on either cases or variables, date
functions and random number generating functions.
ENDBOX;
BOX
SIMSTAT also provides an add-in pulldown menu that gives you
an easy access to SIMSTAT add-in programs, script files,
or other utility programs to perform specific analyses or
tasks not available in SIMSTAT. This feature can also be
used to run other programs you use often.
ENDBOX;
QBOX "SIMSTAT currently comes with 2 addin programs";
BOX
SIMCALC is a probability calculator that can
compute probabilities and confidence intervals
from user supplied statistics.
ENDBOX;
BOX
SIM2NL is an interface program that integrates
NONLIN 3.0, a powerful nonlinear multiple
regression program into SIMSTAT. This program
allows among other things to enter the form
of equation to fit using normal algebric notation
and includes an extensive library of functions
including trig, exponential, log, Bessel, elliptic
integrals, and many others. (see NONLIN.INF for
further information)
ENDBOX;
BOX
The registered version also includes another addin:
SIM2LOG which allows to integrate LOGISTIC v3.11,
a logistic multiple regression program within SIMSTAT.
ENDBOX;
BOX
A special addin program called Banner v1.0 is also available
to registered user for $15. This addin program enables you
to produce presentation quality tables and banners. For more
information about this addin program, select the BANNER v1.0
option from the main menu.
ENDBOX;
BOX
To learn more about the bootstrap resampling technique,
the new features introduced in version 3.5, or the
registration benifits, select the appropriate item from
the demo main menu.
ENDBOX;
GOTO MAINMENU;
************************** BOOTSTRAP *************************;
:BOOT;
BOX
Bootstrap is a resampling technique invented by Bradley
Efron in 1977 which uses the computer's power to alleviate
present limitations of statistical theories.
ENDBOX;
BOX
Bootstrap simulation is a resampling technique whereby initial
sample subjects are treated as if they constitute the population
under study. By replicating those data an infinite number of
times, we then draw at random from that population a large number
of samples, each the same size as the original sample.
ENDBOX;
BOX
By computing, for every bootstrap sample, a statistical estimator of
interest (such as a mean or a correlation between two variables),
this resampling procedure recreates an empirical sampling
distribution of this estimator. Let us see an example:
ENDBOX;
BOX;
The SAMPLE.DAT file contains data from the "law school experiment"
presented by Diaconis & Efron (1983). These authors tried to
estimate the sampling distribution of the correlation between
the average undegraduate grade point and the average score on
the law school admission test from a sample of 15 law schools.
ENDBOX;
DATA
LSAT GPA
576 3.39
635 3.30
558 2.81
578 3.03
666 3.44
580 3.07
555 3.00
661 3.43
651 3.36
605 3.13
653 3.12
575 2.74
545 2.76
572 2.88
594 2.96
ENDDATA;
CLS;
LIST ALL;
BOX
Here are the data
obtained from
those 15 schools.
ENDBOX;
BOX
We will now compute a regression analysis with the
GPA (grade point average) scores as the predictor
and the LSAT (law school admission test) score for
the dependent variable.
ENDBOX;
REGRESSION LSAT BY GPA;
BOX
The data obtained yielded a correlation coefficient of .7764
ENDBOX;
BOX
To estimate the accuracy of this statistic, we will now randomly
select 15 cases from the original sample and compute for each
bootstrap sample a new correlation coefficient. In a given sample,
some schools may be selected more than once whereas others may not
be included, thereby producing a different correlation coefficient.
ENDBOX;
FULL /SAMPLING = 1
/REGRESSION LSAT BY GPA;
BOX
Look at the coefficient of correlation above. You will
notice that the value is different from the one obtained
from the original sample.
ENDBOX;
BOX
We will now perform 3 other regression analyses on
different bootstrap samples, pausing after each
analysis. Look carefully at the values obtained.
ENDBOX;
FULL /SAMPLING = 1
/REGRESSION LSAT BY GPA;
PAUSE;
FULL /SAMPLING = 1
/REGRESSION LSAT BY GPA;
PAUSE;
FULL /SAMPLING = 1
/REGRESSION LSAT BY GPA;
PAUSE;
BOX
If we keep the values obtained for each bootstrap sample
and look at their distribution, we obtain a fairly good
approximation of the sampling variability of that estimator.
Let's look at the distribution of the correlation coefficents
of 250 bootstrap samples.
ENDBOX;
GRAPH /TITLE "Distribution of 250 correlations";
BOOTSTRAP2 LSAT BY GPA
/SAMPLING 250 DESC PTILES=10 INTERVAL=80
/HISTO=G NBAR=40 MIN=0 INC=0.025;
BOX
SIMSTAT also provides various information on the estimators
such as descriptive statistics, percentile tables, and
confidence intervals.
ENDBOX;
BOX
The main advantage of the bootstrap approach over
traditional methods is that the distribution is not
mathematically estimated, but is empirically
constructed on the bases of all of the characteristics
of the original data distribution, including the ones
generally considered contaminating factors
(e.g., skewness, ceiling effect, outliers, etc.).
ENDBOX;
BOX
It can also be useful for describing the sampling
variability of estimators for which the sampling
properties are unknow or unavailable.
ENDBOX;
BOX
For example, boostrap simulation can be used to
obtain a standard deviation of a median even
if there are no mathematical formula available
to compute such a measure.
ENDBOX;
BOX
Typical applications of bootstrap resampling include:
■ computation of nonparametric estimates of sampling distributions.
■ assessment of the stability of statistical models.
■ construction of nonparametric confidence intervals.
ENDBOX;
BOX
SIMSTAT provides bootstrap analysis for seven descriptive
estimators of a single variable and twenty estimators
involving two variables. Those estimators are:
One variable estimators:
■ Mean
■ Median
■ Variance
■ Standard deviation
■ Standard error
■ Skewness
■ Kurtosis
ENDBOX;
BOX
Two-variable estimators:
■ Kendall's tau-a and b
■ Kendall-Stuart's tau-c
■ Symmetric and asymmetric Somers' d
■ Goodman-Kruskal's gamma
■ Student's t and F
■ Pearson's r
■ Spearman's rho
■ Regression slope and intercept
■ Mann-Whitney's U
■ Wilcoxon's W
■ Difference between means
■ Difference between variances
■ Sign test
■ Kruskal-Wallis ANOVA
■ Median test
ENDBOX;
BOX
The FULL ANALYSIS bootstrap procedure also allows you
to perform almost any statistical analysis available
in SIMSTAT on successive bootstrap samples.
ENDBOX;
BOX
For example, it is possible to do a frequency analysis,
a crosstabulation or a multiple regression on each
bootstrap sample. Specific statistics can then be
extracted from the listing file with the use of a text
editor then stored in a new data file for further analysis.
ENDBOX;
BOX
Bootstrap resampling analysis usually involves redrawn
samples of the same size as the original one. SIMSTAT
offers the possibility to modify the dimension of the
bootstrap samples, thus allowing users to compare
estimator distributions obtained from using different
sample sizes. You can set bootstrap simulations
involving sample sizes that range from 2 to 20,000
observations.
ENDBOX;
BOX
The possibility of comparing various estimator distributions
obtained for different sample sizes can prove useful in
planning research by allowing the researcher to determine
the sample size needed to achieve a desired precision level.
ENDBOX;
BOX
It can also be used for power estimation, allowing comparison
of the power attained using various estimators and/or sample
sizes. Researchers thus have an empirical basis for choosing
between two different statistical strategies. In addition,
unlike standard approaches to power estimation which rely
on numerous assumptions, including normal data distribution,
bootstrap power estimates make no distribution assumptions.
ENDBOX;
BOX
As a teaching tool, bootstrap simulation would be effective
in illustrating to new statistics students concepts such as
sampling theory or central limit theorem. It would provide
a simulation of the sampling process of an experiment,
allowing the students to visualize the sampling variability
of given estimators.
ENDBOX;
BOX
By increasing or decreasing sample size, the student can
observe how these changes affect the variability of
estimators or the statistical power of an experiment.
Additionally, bootstrap would be effective in demonstrating
how outliers can affect estimation and how data
transformation can improve population estimates.
ENDBOX;
BOX
For example, it may be possible to perform successive
bootstrap simulations of a variable's mean while increasing
the sample size for each simulation and ask students
to describe what happens.
ENDBOX;
BOX
Let's do it on the GPA variable with
successive sample sizes of 2, 10 and 30.
ENDBOX;
GRAPH /TITLE "Distribution of means (n = 2)" SCALEY (0, 100);
BOOTSTRAP1 GPA
/MEAN SIZE=2 SAMPLING=250 HISTO=G NBAR=40 MIN=2.5 INC=0.025;
GRAPH /TITLE "Distribution of means (n = 10)" SCALEY (0, 100);
BOOTSTRAP1 GPA
/MEAN SIZE=10 SAMPLING=250 HISTO=G NBAR=40 MIN=2.5 INC=0.025;
GRAPH /TITLE "Distribution of means (n = 30)" SCALEY (0, 100);
BOOTSTRAP1 GPA
/MEAN SIZE=30 SAMPLING=250 HISTO=G NBAR=40 MIN=2.5 INC=0.025;
:USER_TEST;
IBOX ROW 4 COLOR BLACK ON LIGHTGRAY;
You may now try it yourself by entering
your own sample size. Enter a number
between 1 and 300. Or enter 0 to end.
ENDBOX;
GETNUM $NUM0 "Sample size : " MIN=0 MAX=300 LEN=3;
IF $NUM0 = 0 THEN GOTO END_TEST;
GRAPH /TITLE "Distribution of means (n = $NUM0)" SCALEY (0, 100);
BOOTSTRAP1 GPA
/MEAN SIZE=$NUM0 SAMPLING=250 HISTO=G NBAR=40 MIN=2.5 INC=0.025;
GOTO USER_TEST;
:END_TEST;
BOX
We have presented here only a glimpse of the bootstrap
capabilities. To learn more about this new technique,
here are some good introductory references to bootstrap;
Diaconis, P., & Efron, B. (1983). Computer-intensive methods
in statistics. SCIENTIFIC AMERICAN, 248(5), 116-130.
Efron, B., & Tibshirani, R.J. (1993). An introduction to the
bootstrap. New-York: Chapman & Hall.
Mooney, C.Z. & Duval, R.D. (1993). Bootstrapping: A nonpara-
metric approach to statistical inference. Beverly Hill:
Sage Publication.
ENDBOX;
GRAPH /TITLE "";
GOTO MAINMENU;
****************************** NEW FEATURES ****************************;
:NEW;
OPEN SAMPLE.SPS;
CLS;
BOX
Simstat v3.5 provides many improvements
to version v3.0. Let's take a quick look
at all those new features.
ENDBOX;
BOX
Many new types of graph have been added
to the numerous ones already available.
ENDBOX;
BOX
For instance, the FREQUENCY command allows
you to display a normal probability plot
of a variable in order to verify whether the
variable is normally distributed.
ENDBOX;
FREQUENCY AGGRESS /PPLOT;
CLS;
BOX
The ONEWAY command which performs a one-way
analysis of variance now includes 3 types
of graphs that allow you to compare the
means of the various groups.
ENDBOX;
BOX
a bar chart representing the mean of every group with
an option to display an error bar representing either
the standard deviation, the standard error or a user
defined confidence interval;
ENDBOX;
ONEWAY AGGRESS BY SIBLING /BARCHART;
ONEWAY AGGRESS BY SIBLING /BARCHART CIBAR;
BOX
an error bar diagram which provide another way
to represent the same kind of information;
ENDBOX;
ONEWAY AGGRESS BY AGE / ERRORCHART;
CLS;
BOX
and a deviation barchart which expresses the deviation
of the mean of each group from the grand mean.
ENDBOX;
ONEWAY AGGRESS BY AGE /DEVCHART;
CLS;
BOX
Beside the new barchart and the error bar
diagram, the T-TEST command also produces
a DUAL HISTOGRAM which allows you to inspect
in detail the data distribution in both groups.
ENDBOX;
QBOX "This type of graph can either be displayed horizontally";
T-TEST AGGRESS BY SEX
/GROUP VALUE (1 2) HISTO NORMAL;
CLS;
QBOX "or vertically";
T-TEST AGGRESS BY SEX
/GROUP VALUE (1 2) HISTO VERTICAL;
CLS;
BOX
A scatterplot matrix can also be obtained from the
CORRELATION command in order to graphically examine
the relationships between numerous variables.
ENDBOX;
CORRELATION NO AGE AGGRESS HOURSTV /XYPLOT TREND;
BOX
Finally, an option to display a normal
probability plot of residual values has
been added to the REGRESSION, MULTIPLE
REGRESSION, and GLM ANOVA/ANCOVA commands.
ENDBOX;
BOX
Here is an example of a normal probability plot of
residual values produced by the REGRESSION command.
ENDBOX;
REGRESSION AGGRESS BY HOURSTV /PPLOT;
CLS;
BOX
Beside all those new graphs, you will find some
minor improvements such as the implementation
of the Levene's test of homogenity of variance
in the T-TEST procedure.
ENDBOX;
CLS;
BOX
Also, when performing a linear, nonlinear or multiple
regression, or a multi-way anova/ancova, predicted and
residual values can now be saved in a separate file
along with the values of all the selected variables.
ENDBOX;
BOX
However, the most important new feature is without any
doubt the implementation of a powerful command language.
To learn more about this language, select the fourth option
of the demonstration main menu.
ENDBOX;
GOTO MAINMENU;
**************** COMMAND LANGUAGE **********************;
:LANGUAGE;
BOX
While SIMSTAT pulldown menus and open panels allow you to
quickly and easily do your analyses, the new command language
can increase your productivity when you need to:
■ keep track of all the analyses performed,
■ provide to someone else a description of what you did,
■ automate statistical processis of your data files,
■ quickly perform similar analyses on different data files.
ENDBOX;
BOX
For example, the following commands open a data file named SAMPLE.SPS,
select all the subjects older than 8 years old, set the listing file
to C:\DATA\FIRST.LIS, open this file for input and finally perform a
t-test analysis to compare the level of aggressivity of male and female
subjects, with dual histogram.
OPEN SAMPLE.SPS;
SELECT AGE > 8;
SET LISTING C:\DATA\FIRST.LIS;
SET DISK ON;
T-TEST AGGRESS BY SEX /GROUP VALUE (1 2) HISTO;
ENDBOX;
BOX
Special features of this command language also allow you
to write interactive tutorials, demonstration programs
or even simple applications to be used by someone else
who is not familiar with SIMSTAT.
ENDBOX;
BOX
Those special commands allow you to display
textual information (just like this screen),
ENDBOX;
BOX
wait for a specific key to be pressed, allowing
you to create menus or question tests like...
ENDBOX;
:QUESTION;
IBOX ROW 6
What does the probability value stands for?
ENDBOX;
BOX KEY=ABCDE ROW 12
A) The probability that the null hypothesis is true
B) The probability that the null hypothesis is false
C) The probability of the data given the null hypothesis
D) The probability of the null hypothesis given the data
E) None of the above
ENDBOX;
IF $ANSWER = 'C' THEN GOTO RIGHT;
QBOX "Wrong";
GOTO QUESTION;
:RIGHT;
QBOX "Right";
IBOX ROW 5;
construct a bouncing bar menu
ENDBOX;
MENU
&First item
&Second item
&Third item
ENDMENU;
QBOX "play sound or music";
SOUND F=400 D=0;
QBOX "WELCOME TO SIMSTAT V3.5" ROW 5 COL 2 DELAY 300 COLOR WHITE ON RED;
SOUND F=600 D=0;
QBOX "WELCOME TO SIMSTAT V3.5" ROW 6 COL 6 DELAY 300 COLOR WHITE ON BROWN;
SOUND F=800 D=0;
QBOX "WELCOME TO SIMSTAT V3.5" ROW 7 COL 10 DELAY 300 COLOR WHITE ON GREEN;
SOUND F=1000 D=0;
QBOX "WELCOME TO SIMSTAT V3.5" ROW 8 COL 14 DELAY 300 COLOR WHITE ON BLACK;
SOUND F=1200 D=0;
QBOX "WELCOME TO SIMSTAT V3.5" ROW 9 COL 18 DELAY 300 COLOR WHITE ON CYAN;
SOUND F=1600 D=0;
QBOX "WELCOME TO SIMSTAT V3.5" ROW 10 COL 22 DELAY 600 COLOR BLACK ON LIGHTGRAY;
SOUND OFF;
IBOX ROW 4
It is also possible to ask the user to enter
a numeric or alphanumeric values that can be
inserted almost anywhere in the command file.
ENDBOX;
GETSTR "What is your first name? " $STR1 LEN 15 CLEAR;
QBOX "Thank you $STR1!";
CLS;
GETNUM "2 + 2 = " $NUM1 LEN 3 CLEAR;
IF $NUM1 = 4 THEN
QBOX "Very good $STR1!";
IF $NUM1 <> 4 THEN
QBOX "Wrong answer $STR1!" COLOR WHITE ON RED;
BOX
Option panels can also be invoked prior to
the computation of an analysis to allow
the user to specify the various options.
ENDBOX;
OPEN SAMPLE.SPS;
BREAKDOWN AGGRESS BY SEX
/PANEL;
DELAY 300;
CLS;
BOX
The command language also allow you to run external
program, perform conditional operation in response to
some conditions or input from the user;
ENDBOX;
BOX
SIMSTAT script files (.SCR) are plain text files
and can be created and edited from within SIMSTAT
or using almost any word processor.
ENDBOX;
BOX
They can be executed from the DOS prompt or run
from within SIMSTAT. They can also be added to
the ADDIN pulldown menu.
ENDBOX;
BOX
A RECORD script feature can automatically generate
proper commands corresponding to the actions you
take using the menus and options panels.
ENDBOX;
BOX
This demonstration program was entirely written using
this new command language. To learn more about it,
read the COMMAND.DOC or examine the content of the
DEMO.SCR file.
ENDBOX;
GOTO MAINMENU;
*********************** ORDER ************************;
:ORDER;
BOX
Registering your copy of SIMSTAT will provide you many benefits
such as:
■ registered version without reminder screens;
■ 170-page manual;
■ minimum of 3 months of free support;
■ free upgrade to the next version;
■ substantial discounts on further upgrades;
■ the ability to edit the title printed on each page;
■ batch program compiler/encrypter;
■ ability to run SIMCALC as a standalone application;
■ SIM2NL addin program to run Phillip Sherrod's NONLIN
from within SIMSTAT;
■ SIM2LOG addin program to run G.E. Dallal's LOGISTIC
from within SIMSTAT (the logistic regression program
is also included on the distribution disk).
ENDBOX;
BOX
For check or money order registration,
send your order to:
Normand Peladeau
5000 Adam Street
Montreal, QC
CANADA, H1V 1W5
Voice/Fax: (514) 899-1672
ENDBOX;
BOX
You can also register SIMSTAT with your credit card (MasterCard,
Visa, AmEx, Discover) from Public (software) Library:
Phone: 1-800-242-4775 (from overseas: 713-524-6394)
FAX: 713-524-6398
CompuServe: 71355,470
Internet: 71355.470@compuserve.com
Mail: Public (Software) Library,
P.O. Box 35705,
Houston, TX
USA 77235-5705.
To register with CIS's Software Registration Service: enter GO SWREG
at the ! prompt and follow the menus. SIMSTAT registration ID is 3339.
SIMSTAT with BANNER v1.0 registration ID is 3562.
ENDBOX;
GOTO MAINMENU;
:BANNER;
CALL BANNER.SCR;
GOTO MAINMENU;