WinCross Frequently Asked Questions


APPLICATION LIMITS

Q: What is the logic limit in WinCross?

 
  Q: What are the record/file size limits in WinCross?
Q: What if my logic requires more than 480 characters?
 
  Q: How many banner columns can I have?
Q: How many rows can be displayed on a single table?   Q: What are “reserved” WinCross variable names?

GENERAL

Q: What are the general rules for rounding in WinCross?
 
  Q: What is the difference between a table mean and row mean?
Q: How can I look up an interview respondent who may have provided “dirty” data?
 
  Q: When would I want to use indexing?
Q: How do you select only certain tables to run?   Q: How do I write output to Excel?
Q: Why won't my SPSS file open in WinCross?   Q: How is weighting in WinCross handled?
Q: Why doesn't WinCross recognize my variable names?   Q: How do I create a summary table using SPSS data?
Q: How do I create a summary table using ASCII or non-SPSS variable data?    

STATISTICAL TESTING

Q: What are the critical values for the confidence levels used for statistical testing?
 
  Q: What are the formulas WinCross uses for its statistical calculations?
 
Q: When doing significance testing table lookups, what type of test does WinCross use?
 
  Q: Why don’t my statistical tests appear on my tables?
Q: Although I have statistics set up to run in my banner, they are not appearing on my tables. What am I doing wrong?
 
  Q: How do I interpret the One-way ANOVA column indicators on my WinCross table?
Q: How is SPSS significance testing with weighted means biased - download article.
 
  Q: Which WinCross Statistical/ Significance test should I use?
Q: What special considerations are there for volumetric Z-test for proportions?

SAMPLE BALANCING

Q: What is Sample Balancing? Is it the same as rim weighting? Are negative weights possible?

SAMPLE SIZES

Q: How do I determine sample sizes for monadic and paired comparison studies?

WEIGHTED SIGNIFICANCE TESTING DOCUMENTS

Weighted Standard Error and its impact on Significance Testing (WinCross vs. Quantum & SPSS)
 
  A Simulation Comparison of WinCross, SPSS, and Mentor Procedures for Estimating the Variance of a Weighted Mean
Alternative Approaches to Significance Testing with Weighted Means    An Analysis of WinCross, SPSS, and Mentor Procedures for Estimating the Variance of a Weighted Mean

ERROR MESSAGES

Q: Why do I get a “Field width does not match code values” error?    

APPLICATION LIMITS

Q: What is the logic limit in WinCross?

A: 480 characters is the limit when entering logic in a banner, filter or row.  

Back to top


Q: What if my logic requires more than 480 characters?

A: Use the ASSIGN statement in the Glossary to create new variables.  

Back to top


Q: How many rows can be displayed on a single table?

A: 6,000.

Back to top


Q: What are the record/file size limits in WinCross?

A: Limit of 65,536 characters and 32,768 variables.

Back to top


Q: How many banner columns can I have?

A: You can have a maximum of 255 banner columns per banner.

 

Back to top


Q: What are “reserved” WinCross variable names?

A: WinCross allows for the use of variable names. However, some variable names are ineligible in that they are reserved for the exclusive use of WinCross.

Variable names cannot use spaces, commas or WinCross-reserved words such as NET, MEAN, or SUB.

Please refer to Appendix "D" in the WinCross User Guide for a list of reserved words and letters.

Back to top


GENERAL

Q: What are the general rules for rounding in WinCross?

A: You can specify how you want WinCross to round percents and statistics in Setup|Job Settings|Rounding.

There are two options for rounding. The 1st option is to round up values over 5 and round down values under 5. For rounding equal to 5, round to the nearest even number. The 2nd option is to always round up values of 5 or more and round down values less than 5.

Rounding in WinCross is based upon three general rules:

Rule 1 - If the remainder beyond the last digit to be reported is less than 5, drop the last digit. Rounding to one decimal place, the number 5.3467 becomes 5.3.

Rule 2 - If the remainder is greater than 5, increase the final digit by 1.  The number 5.798 becomes 5.8 if rounding to 1 digit.

Rule 3 - To prevent rounding bias, if the remainder is exactly 5, then round the last digit to the closest even number.  Thus the number 3.55 (rounded to 1 digit) would be 3.6 (rounding up) and the number 6.450 would round to 6.4 (rounding down) if rounding to 1 decimal.  When the number to the left of the 5 is even, no rounding occurs.  For example, 4.225 would become 4.22.  When the number to the left of 5 is odd, rounding will occur.  For example, 4.215 would become 4.22.     

See page 12 in Hurlburt, R. (1994) Comprehending Behavioral Statistics, Brooks/Cole, Pacific Grove, CA.  

Back to top


Q: How can I look up an interview respondent who may have provided “dirty” data?

A: You can use the TEST statement in the Glossary to search for respondents that break certain conditions.  

Back to top


Q: How do you select only certain tables to run?

A: Depress and hold the CTRL key while individually clicking on those tables you want to run, or use the Edit Selections button in Run|Tables.  

Back to top


Q: Why won't my SPSS file open in WinCross?

 A: Your file may be an uncompressed SPSS file - WinCross only supports compressed SPSS files. 

Back to top


Q: Why doesn't WinCross recognize my variable names?

A: 1) your data file isn’t open or

     2) you don’t have variable names as the first line of your data file or

     3) your logic variables do not match your data file variables.

Back to top


Q: What is the difference between a table mean and row mean?

A: The difference between a table mean and row mean is based on the type of table you’re creating. A table mean is used where every row in the calculation of the mean has only one value. If the rows represent value ranges, then the # (pound symbol) operator must be used to give the row a single value for calculations.

A row mean is appropriate when rows represent a value range and the mean is to be calculated off the actual distribution of data. In that event, a row is created with the appropriate text and logic showing the value range you want to use. In the Row Options dialog box (Setup|Tables|Row Options) you should select the appropriate statistics (Mean, Standard deviation, Standard error, etc.).

Back to top


Q: When would I want to use indexing?

A: Use indexing to create variables which “stack” or “align” data when derotation of the data would otherwise be necessary.

Example of when to use indexed variables:

You have a series of questions asked for different products or brands. Each product/brand does not have a unique set of columns or variables for the questions asked in the series.

The similar variables are “stacked” in an INDEX table in the glossary. Variables within each stack must have the same field width.

                           I1        I2         I3        I4

IDX (1) INDEX {Q1_A ,Q2_A ,Q3_A ,Q4_A }
IDX (2) INDEX {Q1_B ,Q2_B ,Q3_B ,Q4_B}
IDX (3) INDEX {Q1_C ,Q2_C ,Q3_C ,Q4_C}

(Items in bold are not part of the glossary index table, but are shown to illustrate the logic that will be used for columns and rows.)

I# corresponds to an indexed variable (column).

IDX corresponds to an individual level of indexed variables (row).

These labels can be used in any logic statement in the tables, banners or glossary.

EXAMPLE:

Respondents were asked a series of questions regarding three different products:
 

1st Prod Asked 2nd Prod Asked 3rd Prod Asked
Q1 Q1A Q1B Q1C
Q2 Q2A Q2B Q2C
Q3 Q3A Q3B Q3C
Q4 Q4A Q4B Q4C
Q5 Q5A Q5B Q5C
Q6 Q6A Q6B Q6C

Q1=Product Code (Prod. A=1, Prod. B=2, Prod. C=3, Prod. D=4, Prod. E=5, Prod. F=6)
Q2=First time purchasing this type of product (yes/no)?
Q3=Number of purchases made for this product (actual number in data)
Q4=Customer Service Rep was helpful (yes/no)?
Q5=Dollars spent (actual number in data)
Q6=Fair price (yes/no)?

Later in the questionnaire, respondents were asked to rate two of three companies and state the date of their first purchase from the same two companies rated:
 

1st Company

2nd Company

Q12 Q12A Q12B
Q13 Q13A Q13B
Q14 Q14A Q14B
Q15 Q15A Q15B
Q16 Q16A Q16B

Q12=Store Code (Store A=1, Store B=2, Store C=3)
Q13=Rating scale (Excellent=5, Poor=1)
Q14=Rating scale (Excellent=5, Poor=1)
Q15=Rating scale (Excellent=5, Poor=1)
Q16=Year first purchased from this store

GLOSSARY:

INDEX {Q1A,Q2A,Q3A,Q4A,Q5A,Q6A,Q12A,Q13A,Q14A,Q15A,Q16A}
INDEX {Q1B,Q2B,Q3B,Q4B,Q5B,Q6B,Q12B,Q13B,Q14B,Q15B,Q16B}
INDEX {Q1C,Q2C,Q3C,Q4C,Q5C,Q6C,DUMMY,DUMMY,DUMMY,DUMMY,DUMMY1}
 

I1 corresponds to variables Q1A, Q1B and Q1C (Q1)
I2 corresponds to variables Q2A, Q2B and Q2C (Q2)
I3 corresponds to variables Q3A, Q3B and Q3C (Q3)
I4 corresponds to variables Q4A, Q4B and Q4C (Q4)
I5 corresponds to variables Q5A, Q5B and Q5C (Q5)
I6 corresponds to variables Q6A, Q6B and Q6C (Q6)
I7 corresponds to variables Q12A, Q12B and DUMMY (Q12 and a blank field)
I8 corresponds to variables Q13A, Q13B and DUMMY (Q13 and a blank field)
I9 corresponds to variables Q14A, Q14B and DUMMY (Q14 and a blank field)
I10 corresponds to variables Q15A, Q15B and DUMMY (Q15 and a blank field)
I11 corresponds to variables Q16A, Q16B and DUMMY1 (Q16 and a blank field)

TABLE:

Following is an example of the logic for a table for Q3:
 

Stub Text Row Logic
1 - 10 I3 (01-10)
11 - 20 I3 (11-20)
21 - 30 I3 (21-30)

The filter for this table must include I# or IDX (#) in the logic:

Following are examples of filters to use in place of TN (all respondents) in a table with index logic:
 

I1 (1-6) Corresponds to the codes for all 6 products in Q1
IDX (1-3) Corresponds to all 3 rows in the INDEX table set up in the glossary

Following is an example of a filter used to base the table to one product only:
 

I1(1) Filters the table to only include Prod. A

BANNER:

Example of the logic for a banner for Q1:
 

PROD A PROD B PROD C PROD D PROD E PROD F
I1 (1) I1 (2)  I1 (3) I1 (4) I1 (5) I1 (6)

COMPLEX LOGIC:

Indexed variables & non-indexed variables may be combined in logic statements in any order.

Example: I1 (3) & Q20 (2,4) can also be written as:       Q20 (2,4) & I1 (3)

SCAN:

Scan logic may be used with Indexed variables to scan across more than one consecutive indexed variable. The scan will scan columns or variables in the data, not index columns, .

Example: I8 S3(5)—Corresponds to all “Excellent” ratings in Q13, Q14 and Q15

(Since WinCross cannot “net” a respondent’s data when using the INDEX option, a respondent answering “Excellent” to Q13, Q14 and Q15 in the above example would count as 3 in the “Excellent” row frequency.)

REPEAT:

The REPEAT feature can be used in the index table. This is a form of shorthand only. It does not reduce the number of index table columns, nor does it in any way combine or relate the variables to one another. A string of three variables designated in the index table as “Q21A R3” would still be three separate variables. If you wanted to include all three variables in a table, you would use the SCAN feature (see above example). The variables to be designated by the REPEAT feature must be of equal width.

The above index table example could also be written as:

INDEX {Q1A R2,Q3A,Q4A,Q5A,Q6A,Q12A R4,Q16A}
INDEX {Q1B R2,Q3B,Q4B,Q5B,Q6B,Q12B R4,Q16B}
INDEX {Q1C R2,Q3C,Q4C,Q5C,Q6C,DUMMY R4,DUMMY1}

Variables Q1A and Q2A are still designated as “I1” and “I2” just as variable Q3A is still designated as “I3.”

Note: When more than one series of questions needs to be indexed (whether the next series relates to the same products or another variable altogether), all variables must be incorporated into one index table in the glossary. In the above example, Q2 through Q6 (I2 through I6) relate to the product from Q1 (I1). However, Q13 through Q16 (I8 through I11) relate to the store from Q12 (I7).

Because all index rows (IDX 1 through IDX 3 in the above example) must have the same number of variables, “dummy,” or empty, variables must be placed wherever there is no applicable variable. In the above example, Q12 through Q16 are only asked for two stores for each respondent. The index table has three rows, however, because each respondent was asked Q1 through Q6 for three products. Hence, blank variables are used to fill the last row for I7 through I11 (variables DUMMY, DUMMY R4 and DUMMY1).

When dummy variables must be used, it is necessary to either use existing variables that are blank throughout your data file or to add length or a record to your data file.
 

GLOSSARY: ADDLEN 50 = Adds 50 blank columns (for each respondent) to the end of your data file.

GLOSSARY: ADDREC 1 = Adds a full blank record (for each respondent) to your data file.

Back to top


Q: How do I write output to Excel?

A: WinCross tables can be output to Excel. When Run|Tables|Excel Options is selected, WinCross creates an Excel file in addition to the WinCross output file. You also have the option to save your crosstab report using File|Save|Save Report As. Saving to Excel using File|Save|Save Report As produces an image of your WinCross tables and does not contain the actual values the way the Excel report from Run|Tables|Excel Options does.

Other considerations in Run|Tables:

Make sure you have also selected the Create tables option.

Choose the Directory where you want WinCross to write the Excel file(s).

Back to top


Q: How is weighting in WinCross handled?

A: WinCross has the ability to weight your crosstabs. This is done using Setup|Banners|Edit Banner|Weights and must be done after banner creation.

Weights can be created in three ways:

    1. Weight variables can be part of your original data file

    2. You can use glossary statements to create weighted variables

    3. You can specify an actual value for a weight when you set up banners

Weights can be specified with an actual weight value, with the card/column location that holds the weight variable or the variable name for variable data. An actual weight can have up to 12 digits, can be positive or negative and can have decimal places.

     

Back to top


Q: How do I create a summary table using SPSS data?

A: Mean Summary tables are easily created using the Summary of Means option in Setup|Express Tables for SPSS Data. You select the group of variables, choose the code value or range of code values for the rows and WinCross creates the table for you. The Question Title from each variable selected becomes the row text for each row and the Code Value(s) or range of Code Values specified becomes the row logic.

Top/Bottom Box Summary tables can also be easily created using the Summary of Frequencies option in Setup|Express Tables for SPSS Data. You select the group of variables, choose the code value or range of code values for the rows and the code value or range of code values for the percentaging base of each row and WinCross creates the table for you. The Question Title from each variable selected becomes the row text for each row and the Code Value(s) or range of Code Values specified becomes the row logic. WinCross automatically creates two sets of rows, one set to be used for percentaging bases with the Hide row (HR) option specified and the second set of rows is automatically created with the percentaging base specified.

Back to top


Q: How do I create a summary table using ASCII or non-SPSS variable data?

A: To create a mean summary table, your rows will be the question text for the series of tables you are summarizing. Your logic will be the range of values included in the mean calculation.

Note: If using ASCII data, simply replace the variable name in the examples with the card/column location.

Mean Summary

Example—Summarize this series of questions:

Please rate the customer service representative on the following, where:
 3=Low, 2=Medium, 1=High and 4=Don’t know

Knowledge – Q1A
Professionalism – Q1B
Empathy – Q1C

The mean summary table would be defined as follows: 

Knowledge Q1A (1-3)
Professionalism Q1B (1-3)
Empathy Q1C (1-3)

Then choose the mean row option for each of the three rows.

If the scale needs to be reversed, your syntax would be defined as follows:
 

Knowledge Q1A (1-3)#1=3,3=1
Professionalism Q1B (1-3)#1=3,3=1
Empathy Q1C (1-3)#1=3,3=1

Again, choose the mean row option for each of the three rows.

Top Box Summary

If you want to create a summary of just the high score your rows would be defined as follows:
 

Knowledge Q1A (1-4)
Professionalism Q1B (1-4)
Empathy Q1C (1-4)
Knowledge Q1A (1)
Professionalism Q1B (1)
Empathy Q1C (1)

The first three lines are the “basing” lines. The last three lines are the lines indicating respondents saying “High.” The “High” lines need to be percentaged off each corresponding “basing” line using the Setup|Tables|Table Options| Percentage Off option.

Since the first three rows are only used for the calculation, you don’t need to show them on your output and can use the Setup|Tables|Row Options|Hide row option.

Back to top


STATISTICAL TESTING

Q: When doing significance testing table lookups, what type of test does WinCross use?

A: WinCross significance testing is performed using two-tailed tests.

Back to top


Q: Although I have statistics set up to run in my banner, they are not appearing on my tables. What am I doing wrong?

A: You must select Means/Percents/Chi-Square for your table statistics (Setup|Tables|Statistics) as well as for your banner.

Back to top


Q: What are the formulas WinCross uses for its statistical calculations?

 A: The statistical formulas WinCross uses are explained in the Statistical Reference available from the WinCross Help menu.

Back to top


 

Q: Why don’t my statistical tests appear on my tables?

A: The option to include statistical testing has been added at the banner level, but has not been selected at the table level.

Choose Setup|Tables, then select the table to which you want to add statistical testing. Next, choose Statistics, followed by the preferred tests to run. This process only adds statistical testing to the selected table (as opposed to the banner).

Another way to globally add this option to multiple tables is to modify the table statistics. First, choose Setup|Globally Modify Tables|Table Statistics. Choose Means, Percents and/or Chi-Square. Next, select the tables to apply this to (or choose Select All), then choose the Add to existing settings option.

A: Statistical testing has been selected at the table level but not the banner level, or comparison groups in the banner have not been chosen.
Make sure statistical testing has been selected at the banner level. If comparison groups have not been chosen at the banner level, statistical testing selections will not be saved.
          WinCross requires that statistical testing be selected at both the table level AND the banner level.

Back to top


Q: How do I interpret the One-way ANOVA column indicators on my WinCross table output?

A: One-way ANOVA tests are performed on columns of data within each specified WinCross comparison group. WinCross will assign a #1 to the first subset of data within each comparison group where no significant differences were found. Likewise, WinCross will assign a #2 to the next subset of data within the same comparison group where no significant differences were found. This process continues until all subsets of data have been examined for each comparison group.

A column of data can be a member of multiple subsets of data where no significant differences were found. In this case, you will see multiple One-way ANOVA group indicators under that column of the banner on the mean row of your WinCross report. For example, #1#2#3 under a column of data indicates this value is a member of subset 1, subset 2 and subset 3 for that comparison group.


ERROR MESSAGES

Q: Why do I get a “Field width does not match code values” error?

A: There is a value wider than the programmed logic (e.g., Q2 (1-99) or 1/5:2 (1-99) and there is a value higher than 99).

Back to top

                                        


Contact TAG for more information:  Info@AnalyticalGroup.com      


 

Copyright © 2013 The Analytical Group, Inc. All Rights Reserved.