Analysing two surveys together
The worksheet Merging Similar Surveys dealt with using an export to Excel to combine the data of two surveys that were slightly different. This worksheet tells you how to do it within Snap using a database link.
Background
You may need to merge surveys in these situations:
- Surveys run on an annual basis, to analyse trends through two or more years.(The questionnaire might change slightly from year to year.)
- Surveys on one subject but with different focal points (e.g, surveys of users and non-users of a service).
- Surveys conducted for different locations (e.g, e a survey of similar tourist attractions which varies between sites).
- Surveys conducted using two different interviewing methods (e.g, kiosk and web)
This worksheet covers merging two surveys which are slightly different. It assumes that a survey has been conducted of Health Club users over two consecutive years. Q4 in the 2007 survey has one code (Sauna) removed and two codes (Dance Studio and Flotation Tank) added. An question about Ease of Use has been deleted from Q6.
Summary of steps
- Clone the survey that you are going to import into. (In this case, the one with the most variables.)
- Start the database link wizard to import the other survey into the clone.
- Map the variables across the surveys, creating any extra variables needed.
- Import the survey with your specified variable mapping.
- Perform any cross-survey analysis needed.
Step 1: Creating an empty survey to merge into
- Clone the 2007 survey, giving it a useful name such as snHealth Club 06-07. Also clone the raw data. This will give you a new copy of your survey including the existing data. Note how many cases are in the new survey.
- Click the data entry button
to go into data entry mode, and select Database link from the File menu. You will see an empty database link window.
-
Click [New...] to create a new database link.
Step 2: Choosing the import type
- In the Database Linkage Wizard, select Import to Snap Survey as the Linkage Type.
- Click [Next>]. The list of surveys in the current directory appears.
- Select the survey that you wish to merge and click [Next>]. You can browse to a different directory if you need to. You will be asked how you wish to merge the survey.
- Select Append all cases. This adds all the imported cases as new cases at the end of the survey. The other options add more information to existing respondents' details. Click [Next>].
Step 3: Mapping the variables between surveys
When you have selected how to add the new case data, a window opens showing how the variables are mapped between the surveys. The variables in the survey you are importing are on the left. The current survey variables are on the right.
- The first time you import a survey Snap guesses how the variables should be mapped. Since the surveys are similar, the match looks correct, but changes are needed.
- Click each of the right-hand value to display the code values within the question. The values for Q4 are different.
- You must add a new code for the Jacuzzi. Right-click the Solarium code value to show the context menu.
- Click New code. The Create Code dialog appears.
- Type Jacuzzi as the Code Label and click [OK]. A new code is added to your survey. The Jacuzzi responses will be imported into it.
- Adjust the other codes by selecting appropriate variables from the drop-down menu (e.g. in the example above select Restaurant instead of Dance studio for the Restaurant responses to be imported into).
- You also need to add a new variable to the survey. The unmatched variable will be in the list with a blank next to it. Click the blank line and select <Create variable> from the drop-down list.
- A window appears that describes your variable. Check that the details are correct and click [OK].
By default, Snap creates a variable that matches the one you are importing. You might wish to change the Name to fit appropriately into the survey (e.g., in the example, change Q9 to Q6f).
If you are importing a multi-choice variable, the choices are also added automatically. You can change the codes by right-clicking the new code value and selecting from the context menu. You can adjust the mapping by selecting from the drop-down menu.
- Click through all the variables to check that the code values match. When you are content with the survey mapping, click [Next>]. You will be presented with a summary of the your database link.
- Change the default link name if you wish. The type of link will automatically be appended when you finish (in this case [Survey Import]) You can scroll through the survey mapping to see if it is as you expect. Click [Finish] when you are done. The new database link will be added to the Database Linkage dialog.
Step 4: Running the database link
- If the Database Linkage dialog is not already open, open it by selecting File| Database Link...
- Select your new Survey Import link and click [Run]. A dialog appears showing details of the import. Note the number of new cases added (121) in the example, and click [OK].
Creating a year variable
The example given above demonstrates merging two surveys with data from two different years. You can create a new variable to compare one year against the other.
- Create a new derived, single response variable called YEAR.
- Set the label for code 1 to be '2007' and the value to be 'case<=130' (there were 130 cases in the 2007 survey). Set the label for code 2 to be '2006' and the value to be 'true', this will capture all remaining data. You can now perform analysis by year.
You could also set the value for code 2 to be 'case>130 and case<=251' (as there were 121 cases in 2006).
If two surveys relate in a different way, for example, different locations or employee departments, you can create similar derived single variables using the same principles.