Since I started using Pendragon Forms 5.1 in 2007, my biggest gripe has been the lack of integration between Pendragon and statistical software packages like SPSS or STATA. Sure, you can export data to Excel or text files, but a lot of information is lost — namely numeric variable values and value labels.
I’ve posted previously about importing variable labels from Excel to SPSS. That trick saves time, but it does not do enough (and it requires that you start your Pendragon programming in Excel). In this post I discuss a better method for using data stored in Pendragon’s hidden MS Access tables to do the following tasks in STATA:
- create variable labels
- replace string data with numeric values stored in lookup tables
- define value labels
A while back I started using lookup lists in Pendragon Forms. Lookup lists allow you to create a table in Pendragon (MS Access) for each category of response options in your survey. For instance, you can create a lookup list for states where each state is linked to a value, in this case a 2-letter abbreviation. In addition to the “states” list, you might also have a “YesNo” list and an “5ptLikertAgree” list.
Lookup lists are better than popup lists in most situations because you can add new response options once a survey has been frozen. This makes lookup lists a flexible method for presenting response options in a list. These lists, however, are still flawed.
After you create your “states” list, you assign it to all of the questions in your survey that require the respondent to select a state from the list. In Pendragon, choose the “Lookup List” field type and select the correct list as shown below.
Note in the first image that you have the option to “Store the Lookup Values in the lookup field.” This means that you can choose to have the PDA store “PA” rather than “Pennsylvania” when a user selects the Keystone state. The problem is that, when your values are numbers, users cannot easily determine which answer they selected.
For instance, if you choose to store values, and a user selects “Strongly Agree” from the list of response options, the value, let’s say 5, is stored in the database. But this value is also shown on the PDA as the selected response. When you have many lookup lists, it is impossible for enumerators to know what “5″ means.
If you choose not to store the values, the default option is to store the lookup items (e.g., “Pennsylvania”) in the database and on the PDA. This enables the enumerator to review and confirm her selection before proceeding to the next question, but it results in a database filled with string variables.
String variables are easy to recode in SPSS or STATA if you have a codebook that links list items to their values. There is no option to generate a codebook in Pendragon 5.1.
This means that when you export the data from Pendragon to Excel or to a text file, you end up with string variables that need to be manually recoded from a codebook that does not exist.
Until now. Head over to Part 2.
