Advanced calculation options (AskiaAnalyse)

In this topic:

Introduction

A subset of the calculations available in Askia have specific options relating to the calculation, which are provided in addition to the options that can be set for any calculation in the the general calculation properties window.  These are described as advanced calculation options, and they vary according to the calculation type.

To specify any of the options listed on this page, you should:

  1. Find the calculation from the General Tab (see Calculation types)
  2. Right click on the name of the calculation and select Properties
  3. Click the advanced options... button.
Note: If the advanced options.... button is greyed out, this means that the calculation does not have any advanced options to set. 

 

Advanced options for percentage and percentile calculations

The advanced options dialog allows you to set advanced options for the calculation being edited. The available options in this dialog will vary, depending on the type of calculation you are editing.

The options window may may appear in two different formats, according to the calculation being specified, as the options are calculation-specific:

Options%20on%20calculations%201 Options%20on%20calculations%203

Click to enlarge the above images.

The following options are available, according to the context:

COMMAND DESCRIPTION Applies to
Error percentage The level of error used in the calculation, as a percentage.

Standard error, CI on mean (-), CI on mean (+),
Interval (-),  Interval (+),
Use 1 instead If selected, 1 will be used instead of the stated error percentage. Standard error, CI on mean (-), CI on mean (+)
Use standard deviation estimator If selected, the standard deviation estimator will be used. Standard error, CI on mean (-), CI on mean (+),
Percentage This setting determines the level to be used. The percentile calculation returns the number of records that fall below the specified percentage. Percentile

Allow interpolation

If this option is not selected, askiaanalyse will give the first value above the specified percentile. If it is selected, askiaanalyse interpolates the value with the preceding value if it is not an exact match.

 Median, Percentile
 

 

Advanced options for significativity calculations

The options window may may appear in three different formats, according to the calculation being specified, as the options are calculation-specific:

Options%20on%20col%20on%20col%20testing Options%20on%20significativity Options%20on%20Col%20Sig  

 

Click to enlarge the above images.

The following options are available for significativity calculations, according to the context:

COMMAND DESCRIPTION Applies to
Test each column against

A drop-down list of options to select which columns should be compared for significant differences, as follows:

  • All columns: all the columns of the same question or different questions will be compared with each other. For example, a table containing variables for Gender (Male and Female) and Age (Under 24, 25-34, 35 and over) in the columns, the columns would be identified as follows:

A: Male
B: Female
C: Under 24
D: 25-34 years
E: 35 and over

Comparisons would then be made as follows:

A versus B (Male v. Female)
A versus C (Male v. Age under 24)
A versus D (Male v. Age 25-34)
A versus E (Male v. Age 35 and over)
B versus C (Female v. Age under 24)
B versus D (Female v. Age 25-34)
B versus E (Female versus Age 35 and over)
C versus D (Age under 24 v. Age 25-34)
C versus E (Age under 24 v. Age 35 and over)
D versus E (Age 25-34 v. Age 35 and over)

  • All columns of the question: all the columns of the same question will only be compared with each other. For example, for the same Gender and Age questions in the columns, the columns would be identified as follows:

A: Male
B: Female
C: Age under 24
D: Age 25-34 years
E: Age 35 and over

Comparisons would then be made as follows:

A versus B (Male v. Female)
C versus D (Age under 24 v. Age 24-34)
C versus E (Age under 24 v. Age 35 and over)
D versus E (Age 25-34 v. Age 35 and over)

  • All columns of the question and the corresponding question edges: all the columns of the same question will only be compared with each other and with the corresponding edge. For example, for the same Gender placed in the edge and Age question in the columns, the columns would be identified as follows:

A: Male (in the edge)
B: Age under 24 (in the columns)
C: Age 25-34 years (in the columns)
D: Age 35 and over (in the columns)
E: Female (in the edge)
F: Under 24 (in the columns)
G: Age 25-34 years (in the columns)
H: Age 35 and over (in the columns)

Comparisons would then be made as follows:

A versus E (Male v. Female)
B versus C (Age under 25 v. Age 25-34)
B versus D (Age under 25 v. Age 35 and over)
C versus D (Age 25-34 v. Age 35 and over)
B versus F  (Male: Age under 24 v. Female: Age under 24)
C versus G: (Male: Age 25-34 v. Female: Age 24-34)
D versus H:  (Male: Age 35 and over v. Female: Age 35 and over)

  • Previous column only: all the columns of the same question or different questions will be compared to the previous column.

  • Totals only: all the columns of the same question or different questions will be compared to the total column.

  • Specify columns: for column significativity tests, this option allows you to list specific columns to be compared against each other. This is useful if you wish to restrict the test to being performed on specific columns, such as columns that are in some way exceptional. When you select this option, a text field will appear, allowing you to specify which columns you wish to test against each other (see separate entry for Test cols (A:B,C:D-F) below).

  • Specify columns in column profile: for column significativity tests, this option allows you to list specific columns to be compared against each other in the columns tab of the tab definition, instead of in the tab template settings. To do so, you need to right-click the relevant question or questions you want to test in the columns tab, and select col-sig testing....
  • Total and edge total only: for column significativity tests, this option allows you to test all the columns of the question or questions against the total column and edge total.

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, 

Row significativity

 
Using

A drop-down list of options to specify the type of significance test to be used, as follows:

  • classical student test
  • student test using estimator,
  • student test using efficiency coefficient,
  • student test using estimator,
  • efficiency coefficient,
  • Wilcoxon's rank sum test, 
  • ANOVA one way (available for column significativity on numeric calculations)
  • ANOVA with random subject (available for column significativity on numeric calculations).
  • Khi square (available for column on column significativity calculations.
Note: The menu of tests available will vary according to the tests that are relevant to the calculation option being specified. 

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity
On

A drop-down list of options to specify the value on which the significance test will be performed, as follows:

  • For column significativity on closed calculations, you may select the test to be based on the following:
    • frequency
    • mention rate, 
    • frequency by script, 
    • horizontal frequency (using column base
    • horizontal frequency (using row base) or
    • Script (vertical %) 
  • For   column significativity on numeric calculations you may select the test to be based on either:
    • mean or
    • mean by script.  For this option, you can define a script (by clicking script... in the calculation properties), thereby producing a script calculation (mean) with column significativity.

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity
Test against total column If this checkbox option is selected, the total column becomes a column like any other for the purposes of the calculation.

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity
Test against edge total column If this checkbox option is selected, significance testing will be carried out appropriately on edge column groups (e.g. ABCDEF, ABGHI, AJKLMN, AJOPQ), without the need for you to manually specify the columns.

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity
Column selection

Select from the following options for which column should be used as the comparison column:

  • Previous column
  • First column
  • Specify column - when selected, the number of the column to use should be entered in the field Column Index, immediately below this field.
 Col on col significativity
Column index Use in association with Column selection (see separate entry, above) to determine which column should be used as the significativity test comparison column. Col on col significativity
Display minus if under If this checkbox option is selected, a minus sign is shown if the significancy goes down. This can be used in conditional formatting when you test in conjunction with the column before: if there are two letters, the value has significantly gone down, if you have one, it has gone up.

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity
Use student test when degrees of freedom < If this checkbox option is selected, a Student's t-test will be used when the degrees of freedom are less than the amount stated in the adjacent box.

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity
Standard deviation known  

Significance, 

Test Value, Value test (numeric)
Use unweighted base If this option is selected, significativity  or col. significativity will be carried out on weighted % and unweighted counts. All significantivity calculations.
Columns are assumed independent If this option is selected, the individuals belonging to a sub-total will be considered different to those present in the category grouped in the same sub-total.  
Count threshold The minimum count to be taken into account in a cell.

Significance,

Test Value, Value test (numeric),

Column significativity on closed,

Column significativity on numeric,

Col sig on closed, Row significativity
Base threshold The minimum base that must be met before column sig. testing is displayed. By default, the minimum base is 0. This option affects closed and numeric questions. Same as for Count threshold, above
High significativity (%) The percentage at which values are to be regarded as highly significant. Same as for Count threshold, above
Normal significativity (%) The percentage at which values are to be regarded as of normal significance. Same as for Count threshold, above
Low significativity (%) The percentage at which values are to be regarded as of low significance. Same as for Count threshold, above
Display

Select the significativity value to show from the following options:

  • Probability
  • Sigma
 Col on col significativity
Display "A+" Mark highly significant values with A+.

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity
Display "A" Mark values of medium significance with A. Same as for Display "A+", above.
Display "a" Mark values of low significance with a. Same as for Display "A+", above.
Test cols (A:B,C:D-F)

A text field that allows you to list specific columns to be compared against each other. Enter the letters or numbers of the columns you want to test. Separate each test with a comma, and use a colon to separate the columns to be compared within a test. Examples:

  • To test AvsB and DvsE only, enter A:B,D:E or 1:2,4:5

  • To test A against B,C,D,E,F: enter A:B-F

  • To test A,B,C against A,B,C: enter A-C:A-C or A-C

You can display the list of columns being compared in your table (e.g. the footer), by entering the keyword ColSig in the appropriate field.

Only available when Specify columns is selected from the Test each column against drop-down list (see separate entry above).  Applies to:

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity

Sig message (use ??sig??)

Defines a message which indicates at which level the columns have been tested during col significativity testing. The token ??sig?? can then be used to display this message. Select a message in the drop-down list, or enter your own. The items you can place in the message are:

  • p1 = high %

  • p2 = normal %

For example, if the message is "Columns are tested at ??p1??", p1 will be replaced by the high significativity value.

  • p3 = low %

  • invp1 = 100 - high %

  • invp2 = 100 - normal %

  • invp3 = 100 - low %

  • p1_1 = high proba ( 0-1)

  • p1_2 = normal proba ( 0-1)

  • p1_3 = low proba ( 0-1)

  • invp1_1 = 1 - high proba ( 0-1)

  • invp1_2 = 1- normal proba ( 0-1)

  • invp1_3 = 1- low proba ( 0-1)

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity
Prefix / suffix letters Delimiting characters to present immediately before or after significance letters, such as parentheses "(", ")" or brackets "[", "]". Use these to separate multi-character column-sig strings. For example, if "(" and ")" are specified as the prefix and suffix, a string specifying "YM;YF;OM;OF" would be presented as "(YM)(OF)" when the 1st and 3rd columns in sequence are significant. 

Column significativity on closed, 

Column significativity on numeric, 

Col sig on closed, Row significativity

 

 

Advanced options for average number of responses calculations

The following advanced options are available for average number of responses calculations:

COMMAND DESCRIPTION APPLIES TO
Responses selection

Determines which responses are included in the average calculation.

  • Responses with normal base type

  • Responses with positive factors

  • All responses

  average number of responses 
Base selection

Determines which base is used for the average calculation.

  • Same as universe

  • Same as response selection

  • Responses with normal base type

  • Responses with positive factors

  • All responses

  average number of responses 

 

 

Proportions are calculated as follows:

p1= Proportion 1 observed in the sample
n1=Sample size  

p2= Proportion 2 observed in the sample
n2=Sample size

tα= 90% = 1.65
tα= 95% = 1.96
tα= 99% = 2.576

We calculate 'D', which follows a normal mathematical expectation law p1-p2, and standard deviation sd=squared ROOT((p1*(1-p1))/n1 + (p2*(1-p2))/n2),

(P1-P2)

D= --------------------------------------------------------

         P1*(1-P1)                  P2 * (1 - P2)

 Root (-----------------     +   ------------------- )

            N1                          N2

 

=> if abs(D)> tα = Significant difference

 

Means are calculated as follows:

m1= Mean 1 observed in the sample
sd1= Standard deviation 1 observed in the sample
n1=Sample size

m2= mean 2 observed in the sample
sd2= Standard deviation 2 observed in the sample
n2=Sample size

tα= 90% = 1.65
tα= 95% = 1.96
tα= 99% = 2.576

We calculate 'D', which follows a normal mathematical expectation law m1-m2=0, and standard deviation sd=root( sd1 ²/ n1 + sd2²/n2),

(m1-m2)

D= --------------------------------

 squared root( sd1 ²/ n1 + sd2²/n2)

 

=> if abs(D)> tα = Significant difference

Create your own Knowledge Base