IMPORTING DATA :: Excel Data
To import Microsoft Excel data into Interaction!,
click on the 'Browse' button in Step 1 of the New Graph Wizard, and select the
'Excel Files' option from the file type dropdown list in the dialog box that
appears. You will then be able to select your Excel file. After selecting your
file, you can select which worksheet in the file you want to use, and will be
able to specify whether or not the first row in the file contains the names of
the variables in the dataset.
Note that Microsoft Excel must be installed on your computer in order
to import Excel data into Interaction!
Return to help index
IMPORTING DATA :: SPSS Data
To import SPSS data into Interaction!,
click on the 'Browse' button in Step 1 of the New Graph Wizard, and select the
'SPSS Data Files' option from the file type dropdown list in the dialog box
that appears. You will then be able to select your SPSS data file. Importation
of your SPSS data will then be handled automatically by Interaction!
Return to help index
IMPORTING DATA :: Tabdelimited Data
To import a tabdelimited data source into Interaction!,
click on the 'Browse' button in Step 1 of the New Graph Wizard, and select the
'TabDelimited Data File' option from the file type dropdown list in the
dialog box that appears. You will then be able to select your tabdelimited
data file. After selecting your data file, you can specify whether or not the
first row in the file contains the names of the variables in the dataset.
Note that in order to use tabdelimited data with Interaction!,
not only must the unique fields in the data file be separated by a TAB, but the
file must also be a standard ASCII text file with a file extension of '.DAT'.
In addition to these considerations, the rows in the data file must represent
cases, while the columns in the data file must represent variables.
Return to help index
INTERACTION LEVELS :: With a Categorical Moderator
Categorical variables are those variables that can take on a finite
set of values (e.g., eye color, Likertscale data). Interaction!
classifies any variable with between three and ten unique values as
categorical. When the variable chosen as the moderator is categorical,
Interaction! will allow you to either use specific values of the
categorical variable to draw the interaction graph, or to treat the categorical
variable as a continuous variable, in which case the considerations associated
with a continuous moderator apply.
Return to help index
INTERACTION LEVELS :: With a Continuous Moderator
Continuous variables are those variables that can take on a large,
possibly infinite, number of values (e.g., height, distance, temperature).
Interaction! automatically classifies any variable with more than
ten unique values as continuous. When the variable chosen as the moderator is
continuous, Interaction! will provide you with
seven computed values from which to select when drawing the interaction graph.
These values are the mean, +1, +2, and +3 standard deviations above the mean,
and 1, 2, and 3 standard deviations below the mean.
Return to help index
INTERACTION LEVELS :: With a Dichotomous Moderator
Dichotomous variables are those variables whose values can only take
on one of two possible conditions (e.g., yes/no, true/false, male/female). When
the variable chosen as the moderator is dichotomous, Interaction!
will use the two unique values of the moderator to draw the interaction graph.
Note: dichotomous variables must be dummycoded in order to be used with
Interaction!
Return to help index
MISSING VALUES :: Identifying With an Empty Field
If you have missing values in your dataset that are
identified with an empty field, you can indicate this to Interaction!
in Step 2 of the New Graph Wizard. If this option is chosen, Interaction!
will ignore cases whose values are null (i.e., empty) when performing
statistical computations or drawing interaction graphs.
Return to help index
MISSING VALUES :: Identifying With a Number
If you use a specific number in your dataset to identify missing
values, you can indicate this number to Interaction!
in Step 2 of the New Graph Wizard. If this option is chosen, Interaction!
will ignore cases whose values correspond to the specified value when
performing statistical computations or drawing interaction graphs. By default,
the number 999999 is used to indicate missing values when this option is
selected.
Return to help index
MISSING VALUES :: Listwise Deletion of Cases
Interaction! utilizes listwise deletion
when handling missing values. This means that if a case has a missing value for
any of the variables in the model, the case will be not be used in statistical
computation, nor will it be considered when drawing interaction graphs.
Return to help index
VARIABLES :: Covariates
In Interaction!,
covariates are predictor variables that are posited to impact the value of the
dependent variable, but are outside of the scope of the bivariate interaction.
Covariates are optional model components in Interaction!
In the figure above, 'C_{1}' and 'C_{2}'
are covariates. Interaction! can handle any number
of covariates. The covariates are specified in Step 3 of the New Graph Wizard.
Return to help index
VARIABLES :: Dependent
The dependent variable is the variable in the linear model that is
posited to be caused by the set of predictors.
In the figure above, 'Y' is the dependent variable.
Values of this variable are shown along the Yaxis in Interaction!
The dependent variable is specified in Step 3 of the New Graph Wizard.
Return to help index
VARIABLES :: Independent
In Interaction!, the
independent variable is the predictor of primary interest to the researcher,
and is a component of the interaction term. Along with the moderator,
interaction term, and covariates, changes in the independent variable are
posited to cause changes in the dependent variable.
In the figure above, 'X_{1}' is the
independent variable. Values of this variable are shown along the Xaxis in
Interaction! The independent variable is specified
in Step 3 of the New Graph Wizard.
Return to help index
VARIABLES :: Interaction Term
The interaction term is a predictor variable
that is equal to the product of the independent variable and the moderator.
This term is computed automatically by Interaction!
In the figure above, 'X_{1}X_{2}' is
the crossproduct interaction term.
Return to help index
VARIABLES :: Moderator
In Interaction!,
the moderator predictor variable is a component of the interaction term. The
nature of the relationship between the independent variable and the dependent
variable is posited to change based on the value of the moderator variable.
In the figure above, 'X_{2}' is the
moderator variable. Different values of the moderator variable are used to
compute the individual interaction lines drawn on the interaction graph. The
moderator variable is specified in Step 3 of the New Graph Wizard.
Return to help index
