The following options are available when you click analysis options... in the multivariate analysis general tab:
Inactive interviews: This option is available in typology, linear regression and principal components analyses. It determines how the sub-population and inactive sub-population (in the general tab of multivariate analysis) are applied to the analysis.
From active interviews (in the sub-population): Interviews are selected from the specified sub-population, and made inactive.
From unselected interviews: The interviews that are in the active and inactive sub-populations are marked as inactive.
From unselected interviews if they're not active: The interviews that are in the active and inactive sub-populations are marked as active. This is the default setting.
Missing values: This option is available in linear regression and principal components analyses. It determines how missing values are handled.
Replace by mean: Missing values will be replaced by the mean for the respondent in question, and the interview will remain active. This is the default setting.
Replace by zero: Missing values will be replaced by zero, and the interview will remain active.
Make inactive and replace by mean: Missing values will be transformed into the mean for the respondent in question, and the interview will be made inactive.
Make inactive and replace by zero: missing values will be replaced with zero, and the interview will be made inactive.
Remove from analysis: the interview will be removed from active interviews if the respondent has given at least one missing value.
Keep as non response: only the missing value will be converted to DK; other values remain.
Name of the axis: This option is available in principal components analyses. It specifies the text to be used on the output axis.
Name of the interviews: This option is available in principal components analyses. You can use the keyword ??question?? to include the question name.
Typologies to run: Available for typology analyses. Specifies the maximum number of typologies to run in the analysis.
Selection method: This option is available for typology analyses.
Typology with the maximum inter-group inertia: The typology with the greatest inter-group inertia will be selected.
Most frequent repetition: The typology that appears the most often will be selected.
Most frequent then use lowest inertia: The typology that appears the most often will be selected; if there are ties, then inertia is used to select the best match.
Converge when: This option is available for typology analyses.
Stop when inertia variation is below...: Convergence will occur when the inertia variation reaches the threshold specified in inertia variation % below.
Stop when groups are stable: Convergence will occur when the groups are stable.
Inertia variation %: available for typology analyses. This value determines the threshold at which convergence will occur, if converge when (see above) is set to stop when inertia variation is below....
Max number of iterations: this option is available for typology analyses. The default value is 30.
Random seed: this option is available for typology analyses. It determines the seed selection for prototype interviews. The default value is 10.
Sigma method: this option is available for specifities and typology analyses. For details of the formulae used, please see Comparing proportions (askiaanalyse), and the associated topics. The available options are using Khi square, against all other columns with standard deviation unknown and against all other columns with standard deviation known.
Sigma threshold: this option is available for specifities and typology analyses. You can enter a value (e.g. 1.64, 1.96 or 2.58) or a percentage (e.g. 90%, 95% or 99%) by selecting display as percentage.
Sort modalities: This options is available for specifities and typology analyses. If it is selected, items within each group will be sorted by count.
Show significantly low responses: This option is available for specifities and typology analyses. If it is selected, any responses that do not explain the group will be included at the bottom of the results page. Note that the sigma and the significancy percentage will be negative in the output.
You can save your multivariate analysis as a variable, either as a new variable or by updating an existing one. To do so, simply click create variable... in the multivariate analysis general tab.
The following options are available:
Source: select the appropriate item based on the axis (the value of the contribution of each interview to this axis will be saved) and whether you want to save the results or the model.
Shortcut: enter a name for the variable. If you want to save the information to an existing variable, you should enter the name of that variable here.
Caption: enter a caption for the variable.
After: determines the position at which the variable will appear in the variables list (select the variable it will appear after).
Modalities: displays the modalities that the variable will contain.
To save a new variable, click create new. To save the information to an existing variable, click update existing; note that you need to have entered the shortcut of an existing variable in the shortcut field.