Design Studio

Problem:

Description:

New build option added for automated interviews. (R6)

A new check box, "Disallow resumption of completed interviews", was added to the Automated Interviews Build Options dialog box.. When this box is checked, the Show All button on the ACASI/CAPI Pending Interviews dialog box is disabled. This prevents users from resuming and possibly modifying completed interviews.

Build function generated invalid control file when unable to locate bitmap file. (R6)

The Build CAPI/A-CASI function would generate an invalid control file when it failed to locate a bitmap specified in a Pick-one response item. The program displayed an error message, stating that text would be substituted for the missing bitmap, and proceeded to build the file, however, the resulting control file was invalid. The CAPI/A-CASI program would report that it was unable to read the file and terminate.

Fatal error in Build function when alternate skip code defined as "0". (R6)

The CAPI/A-CASI Build function crashed when user defined an alternate skip code as "0". This bug was only in version 2.0 R004. This was fixed so that "0", or any other single digit, does not cause the build step to crash.

Validation function modified to support duplicate variable names in limited circumstances. (R6)

The Validation function was modified to allow duplicate variable names between Data elements and Automatic variables under limited circumstances: the automatic must precede the data element sequentially in the specifications file, and the value type of the automatic must be compatible with the value type of the data element. The following types are compatible: (1) Elapsed time and numeric calculation type automatics are compatible with number, currency, pattern, pick-one, yes/no, gender, and numeric rating scale data elements. (2) Today's date automatics are compatible with Date data elements. (3) Current time automatics are compatible with Time of Day data elements. (4) String expression, mode, language, data file name, and character string automatics are compatible with text data elements. Note that the program issues a warning message when it detects a duplicate name so that users will be alerted in cases where duplication is not intentional.

Options function failed to reject alphabetic code patterns. (R6)

The Options | Data Defaults | Special Codes property page was modified to reject single letter code patterns. For example, in the past, a user could define a skip code as "s". However, the program did not use this value; instead, it filled every skipped response with zero. Now, the program issues an error message if the user enters a single letter in a Code Pattern input field.

Save a Copy As… function did not handle i/o errors properly. (R6)

The Save a Copy As… function did not behave properly when an i/o error occurred. If the copy failed, the program did not issue an error message, and, some cases, the program corrupted the QDS™ file during the save operation.

Design Studio froze during File Recovery. (R6)

Occasionally, the File Recovery function would go into an infinite loop, which required the user to terminate the program via the Windows' Task Manager.

Leading asterisks in Question Text field.

Under certain configurations of Windows, the Design Studio inserted two asterisks at the beginning of each question text field and each response item description field. Each time the user modified an element and then clicked OK, the program inserted two more asterisks. This problem only occurred on systems running Windows NT with Microsoft Office 97 (second release).

Formatted text occasionally disappeared from Question Text field.

When the user applied character formatting - bold, italics, or underline - to question text, clicked OK, and then later re-opened the element, the Question Text field was empty. This problem only occurred when the first character in the formatted string was a foreign character, such as é or ¿.

Fatal error during Validation after adding second language.

This problem occurred when an automatic character string was left blank for the second language.

Validation failed if "No Audio" was used inside a table.

The Validation function issued an error message when a table element contained a wave file name of "No Audio". This was fixed; users can now turn off audio playback for a given element in a table.

Validation did not detect a skip that branched beyond the end of a questionnaire.

The Validation function did not issue an error message for skip elements or edit elements which contained a branch, or a reference to reconciliation elements, which exceeded the length of the questionnaire.

Validation did not detect a table element with the same variable name as a data element outside of a table.

For example, a questionnaire contained a table with categories A, B, and C. A data element inside the table had a variable name of "Q10&C" and a data element outside of the table had a variable name of "Q10A". These were duplicate variables which should have been flagged by the Validation function. This problem sometimes caused the CAPI and A-CASI programs to crash.

Improper skip code defined when .NA disallowed for a given element.

For example, a questionnaire contained a pick-one element with a range of responses from 0 to 9 and all special codes were disallowed. When this item was skipped during an automated interview, a 9 instead of a 99 was assigned to this variable. This was fixed so that the correct number of digits are always used for skip codes. This fix affects the following types of data elements: Pick-one, Time Span, Money, Numbers, Numeric rating scales, and Text.

Improper skip code defined for check-all type data element.

The Design Studio miscalculated the skip code for a check-all type data element with 9 choices. The program defined a skip code of 9, which conflicted with a normal response of 9 (all checked). The program was fixed to use a skip code of "99" whenever the number of choices exceeds 6.

Design Studio did not detect special codes which conflicted with normal response values.

For example, a questionnaire contained a pick-one element with a range of responses from 0 to 9. The program automatically set the special codes to 97, 98, and 99. However, the user was allowed to manually override these and enter 7, 8, and 9. This was changed; the program now issues an error message when this occurs.

Users unable to modify special codes for response card data elements.

For example, a pick-one response card contained a range of values between 1 and 5. The user defined a data element for this response card and set the special code values to 6, 7, and 8. The program appeared to allow this but actually changed them back to 97, 98, and 99 during the Validation step. This was fixed to allow the user to manually override the default values, as long as they don't conflict with the range of normal responses.

Warehouse Manager

Problem:

Description:

Fatal error during value update procedure after reconciliation. (R6)

The Warehouse Manager crashed after the following steps were executed: (1) The user reconciled one version of the questionnaire to another. During this process, a Date variable from the reconciled version was added to the standard version. (2) The user then converted the reconciled cases to the standard version. (3) The user attempted to modify one of the components of the date variable in a converted case, causing the warehouse to crash. This problem was fixed so that composite variables (Date, Time, Time Span, and Check-all) can be modified after being appended to the standard version.

Reconciliation function prompted user multiple time for sub-components of same variable. (R6)

The Warehouse Manager reconciliation process was modified to suppress Discard? message for sub-components of a composite variable. For example, if the reconciled version contained a Date variable named INTDT that was not found in the standard, the program would ask the user is they wished to discard variables INTDT, INTDTY, INTDTM, and INTDTD. Now it only asks the user about INTDT. If the user says "Yes", all four variables are discarded. If the user responds "NO", all four are added to the standard.

Response data sometimes lost during update procedure. (R6)

Response data was lost by the Warehouse Manager when the following steps were executed: (1) The user opened the Value Update dialog for a variable. (2) The user changed response from Normal to "Don't Know", or some other special code, and then changed it back again to "Normal" and selected OK. (So, there was no actual change to the data.) (3) The user selected OK for the entire Interview Details page. (4) The user opened the Value Update dialog box for a different variable in the same case, modified it, and selected OK. (Now, there is an actual change to the data.) (5) The user selected OK for the entire Interview Details page. At this point, the response for the variable that was never really changed was permanently missing from the warehouse file.

Update to Time Span composite did not change sub-component variables. (R6)

The Warehouse Manager's Value Update feature allowed users to modify the value for the main, composite variable for a Time Span element; however, it did not automatically update the sub-components. The program was modified to prevent users from changing a Time Span element this way. The "Normal response" radio button is now disabled for all composite variables, forcing users to update via the sub-components.

During reconciliation, the Warehouse added new variables to the standard version twice instead of once.

For example, assume a warehouse contained two versions; version #1 had the following variables: SUBJECT, AGE, and GENDER. Version #2 had only SUBJECT and GENDER. When reconciling #1 to #2, the Warehouse would display the following prompt, "Variable AGE is not in the standard version. Discard it?". If the user selected No, the warehouse added two new variables to the standard, AGE and AGE1, instead of just AGE.

When reconciling more than one version, the Warehouse added the same variable from different versions repeatedly for each reconciliation, resulting in a lot of unnecessary variables in the standard.

For example, assume a warehouse contained three versions. Version #1 had the following variables: SUBJECT, Q1, and Q2, and version #2 had the same variable names, but the meaning of Q1 had changed. Version #3 was identical to version #2 except for the elimination of a "Not applicable" response for Q2. When reconciling #2 to #1, the Warehouse Manager displayed the "Equivalent?" dialog box for variables Q1, and, since these were different, the user selected No. The Warehouse then generated a new variable for the standard, Q1A. So far, this was correct. However, when reconciling version #3, the Warehouse displayed the same prompt again, not recognizing that Q1 from version #3 was the same as Q1A in the standard version. The result was that the standard ended up with variables Q1A and Q1B for the same question.

The Cancel function on the "Equivalent?" dialog box did not work properly.

During reconciliation, when the user selected Cancel on the "Equivalent?" dialog box, the reconciliation process stopped, but the version being processed was marked as reconciled anyway (preventing the user from reconciling later), and all the variables defined for that version appeared to be missing. This function now restores the variables to their original state, so that the user can re-start reconciliation at a later time.

Text of message during reconciliation unclear.

The message box, "Reconciling this version to the current standard", was changed to display, "Reconciling version #1 to version #2, the current standard.". This was done to clarify which version is being processed when running reconciliation and conversion of interviews in one step.

Revisions to warehouse data caused large increase in file size.

After reconciling versions and converting interviews in a warehouse, the file size would almost double. To address this problem, a new function was added to the warehouse File menu: "Save and Compact". This function makes a copy of the warehouse file eliminating unused portions of the file. When a great number of revisions have been made to a warehouse, such as occur during reconciliation, this function should significantly reduce the file size of the warehouse file.

Fatal error during conversion of interviews.

During conversion of interviews to the standard version, the Warehouse Manager would occasionally display an "Out of memory" error and crash. The problem was triggered by the existence of a variable in a non-standard version which was not in the standard. The warehouse was fixed to properly handle "discarded" variables when converting interviews to the standard.

Fatal error during export of interviews.

When exporting interviews from a large warehouse file, the Warehouse Manager would occasionally run out of dynamic memory and crash. The program was modified to use less dynamic memory.

Fatal error after File | Save performed.

The Warehouse Manager would occasionally crash after performing a File | Save operation where deleted interviews were removed from the file.

Warehouse appeared to freeze during File | Save procedure.

When performing a File | Save operation on a warehouse which removes deleted interviews from the file, the Warehouse Manager displays the following message if it encounters a "disk is full" error: "Cannot write to file — disk is full". The user is presented with two buttons, "Retry" and "Cancel". When the user selected "Retry", the program got stuck in an infinite loop and appeared to freeze. Now, if the user frees up adequate disk space and then selects "Retry", the program should continue writing without any problems.

Warehouse did not report i/o error during File | Save as procedure.

When saving data using the File | Save as... function, the Warehouse Manager did not display an error message when an i/o error, such as insufficient disk space, occurred.

The "Yes to All" function on the "Replace interview?" dialog box did not work.

When copying interviews from a data file to a warehouse, the Warehouse Manager prompts the user with a "Replace interview?" dialog box for all duplicate interviews. The dialog’s "Yes to All" button did not work, causing the prompt to re-appear for each duplicate ID. In addition, the "Date Out" field for replaced interviews was filled with "12/31/69" instead of the current date.

File extension of Interview Comments Log duplicates extension of Data Entry Verification Log.

The file extension for the Warehouse Manager’s Interview Comments Log was the same as that used for QDE verification logs -- ".LOG". This was changed to ".TXT". This was done to prevent the user from mistakenly overwriting one type of log with another.

The Value Update function did not work properly in version 2.0 R002.

If the user attempted to modify a component variable of a date variable, such as year or month, the program crashed with "an invalid page fault" system error. Also, if the user modified two or more variables within the same interview and then selected OK to close the interview details dialog box, the program would lose the data for the modified variables.

CAPI / A-CASI / Data Entry

Problem:

Description:

The Data Entry program did not handle special codes correctly for Check-All elements. (R6)

If a special code was entered for the first field of a Check-All element, the special tool bar buttons (.D, .R, and .N) became disabled and the user was forced to enter the numeric value of the code in the remaining fields. This was fixed so that selecting a special code button or entering a ".D", ".R", or ".N" in the first check box causes the entire set of fields to be automatically filled with the same code.

Data Entry Verification procedure did not treat "09" as a match for "9". (R6)

When verifying data in the Data Entry program, entering a "9" for a field stored as an "09" was flagged as a mismatch.

CAPI/A-CASI/Data Entry did not truncate long interviewer comments. (R6)

The CAPI, A-CASI, and Data Entry programs did not limit the number of characters entered into the Interviewer's Comments dialog box. This caused a problem in the Warehouse Manager when exporting to Microsoft Access, since Access limits field size to 255 characters. This problem was resolved by limiting the number of characters that can be entered into a comments dialog box to 255.

A-CASI did not always save response for numeric rating scale elements. (R6)

If a numeric rating scale element was defined with the Audio Interruptible feature turned off, the program failed to save the user's response in the following situation: (1) The user selected a point on the scale during audio playback and (2) the user selected the Next Question button after audio playback was complete. This was fixed so that the rating scale control is disabled until the audio playback is complete and all responses are saved to the data file.

A-CASI did not always save response for check-all elements. (R6)

If a check-all response was defined with the "If none checked, branch to" option set to "Disallow" and the A-CASI file was built with the "Read choices" audio option turned on, the program failed to save the user's response in the following situation: (1) The user selected one or more choices before the audio playback was complete (2) The user then selected the Next Question button after the audio playback was complete.

PGDN key caused CAPI/A-CASI program to skip current question. (R6)

This problem only occurred in A-CASI when the control file was built with "Support keyboard input" turned on. If the subject pressed the PGDN key, the program would skip the current question, leaving the associated response as "missing", and move to the next question. This was fixed so that the PGDN key only moves ahead to the next element if the question has already been answered and the Next Question button has been enabled.

Escape key terminated CAPI/A-CASI program. (R6)

This problem only occurred in A-CASI when the control file was built with "Support keyboard input" turned on. The subject was able to terminate an interview by pressing the Escape key. This was disabled. The only way for the subject to exit is by pressing Alt-F4 or by completing the interview.

A-CASI did not allow users to cancel Exit Password dialog box. (R6)

The Exit Password dialog box in an A-CASI interview prevents subjects from exiting the A-CASI program upon completion of an interview. However, in the case where the subject mistakenly pressed Alt-F4 before the end of the interview, there was no way to close the Exit Password dialog and resume the interview. This was changed. Now, if the interview is incomplete, pressing the Escape key closes the Exit dialog box and returns control to the current question. If the interview is complete, however, the Exit dialog works as it did before; the program does nothing until the user enters the appropriate password and selects the End button.

Fatal error in CAPI/A-CASI/Data Entry when performing skip. (R6)

The CAPI, A-CASI, and Data Entry programs crashed when attempting to skip a data element that had multiple items checked on its If-Skipped property page and one item was missing a conditional expression.

CAPI/A-CASI did not always branch correctly for pick-one elements. (R6)

If a pick-one element was defined with the "Display in reverse order" option turned on and the "If code zero, branch to" option set to a value other than the default, the program did not branch correctly when the user selected the zero coded response.

Text of "Delete interview" message not clear to all users. (R6)

At the end of a CAPI/A-CASI interview, when user selected "NO" on the "Do you want to save the data for this interview?" message box, the program displayed the following message: "STOP! … Are you sure you wish to continue?". In order to emphasize that the current interview would be deleted, this was changed to "STOP! … Are you sure you want to delete this interview?".

A-CASI program did not support key to activate "Repeat Question". (R6)

If the user built the A-CASI program with "Support keyboard input" turned on, there was no way to activate the "Repeat Question" button via the keyboard. This was fixed so that the F12 key now activates the Repeat Question button.

Numeric values with decimal places saved incorrectly.

Numeric values with two decimal places were not saved correctly to the data file. For example, a value of "1.50" was saved as "1.515152". This problem only occurred in questionnaires containing a DBVALUE statement.

Double-clicks caused the CAPI/A-CASI programs to skip subsequent questions.

For example, if a questionnaire contained two Yes/No questions in a row and the user double-clicked on the No button (or types Alt-N two times in a row), the program processed the second click as a No response to the second question.

Composite variable for time span elements not calculated correctly.

For example, if the user entered 1 month, 1 week, and 1 day as a response, the program stored "3 days" as the combined response instead of "38 days". This was only a problem in the CAPI and Data Entry programs; it was not a problem for A-CASI.

Occasional failure to detect out-of-range errors for time span elements.

For example, if the user entered 90 Years and 9 months as a response for an element which had a maximum value of 90 Years, the program accepted the response, even though the value (after rounding) was stored as 91 years.

A-CASI using incorrect maximum value for time span buttons.

The ACASI program was not using the correct maximum value for time span controls when the components were non-consecutive. For example, a time span element consisted of two units, Months and Days. When the user incremented the Days component using the ">>" or ">" buttons, the program displayed values up to 6 and then wrapped around to 0. The program was fixed to allow the user to select from 0 to 29 for this component.

Inappropriate image used for out-of-range error message.

The bitmap image of a bomb was replaced with a picture of an exclamation point.

Resource leak in CAPI/ACASI module.

Sometimes, during the course of an interview using the ACASI or CAPI module, the text disappeared from the screen and the operating system reported that "A required resource was unavailable". This problem only occurred under Windows 95/98.