Software for surveys

Surveys seem to be everywhere. Unsolicited surveys arrive in the mail (both postal mail and electronic mail), are broadcast on television (as "viewer polls"), and even invade our personal homes as telephone intrusions ("Sir, we noticed that you have bought our products in the past and want to make sure that they meet your needs."). In fact, some unsolicited surveys even arrive with a crisp green dollar bill as an enticement to complete the survey questionnaire and return it to the sender. There are management surveys of employee opinions, business surveys of customer satisfaction, magazine surveys of their readership, and even "feedback" forms to encourage computer users to share their requirements with computer software vendors. Sometimes it seems as though one could spend one day a week just filling out forms and answering questions.

A very common form of survey is a membership survey such as the ones often conducted by professional associations of their constituencies. After all, how will one ever know what the rank and file are up to if they are not surveyed regularly? The first thought of every new committee chair must be to initiate an effort to seek feedback for injection into long range plans, project creation decisions, and mission statements. Without this body of information to provide a communication link to those that such organizations are to serve, how can the relevancy of projects and tasks be guaranteed?

Surveys are, therefore, legitimate and necessary instruments for data collection and compilation without which many organizations and individuals would be working in a vacuum. The challenge is to assure that the content of the survey is relevant, the conducting of the survey is infrequent, and the survey instrument itself is as "user friendly" as possible. Anyone who has led such an initiative can attest to the fact that such activities are "one shot" efforts. Despite the fact that a survey can be repeated periodically, one needs to "get it right the first time." Few intrusions are as annoying as having to fill out for a second time a lengthy survey form that left important questions out of the original effort.

Typical major components of almost all surveys are 1) create a scope, approach, and project plan; 2) create a survey form; 3) distribute the survey form; 4) enter the data received in the responses into a database; 5) perform data analysis; and 6) report the results of the survey. Each of these steps can be aided by computer software of various kinds. However, it becomes very important to make decisions early about which software will be used for each phase of the project and what the overall expected results of the project are to be. One can select different software packages to help with each project task or select an integrated package that will support the entire survey process and data analysis. A brief investigation into the possibilities of information management technology to help conduct surveys seems to be in order.

DATABASE OR WORD PROCESSTNG SOFTWARE?

Most first time surveyors assume that they will be dealing primarily with steps two (create a survey form), three (distribute the survey form), and six (report the results of the survey). Their orientation is to 1) decide what they want the recipient to tell them, 2) get the survey to the people, and 3) record and report the results. Most individuals' first glimmer of the complexities of what they are getting into arises when they attempt to design the form and the questions to be answered. The form design task may seem to be trivial or intimidating, depending on their desktop publishing skills with a word processor. The content of the form may seem challenging when they actually start thinking of all of the questions to be asked and how to ask them.

Suddenly, there is a realization that one must be very comprehensive in the questionnaire, while not boring or overwhelming the person whose opinion is desired. In addition, just creating an attractive and pleasant form that organizes the questions in a meaningful manner is an undertaking unto itself. This forms creation (desktop publishing) task alone can easily consume an entire day, depending on the complexity of the questions and the layout of the form. Questions requiring text responses demand adequate white space on the form for writing or typing in responses. Questions that are multiple choice will benefit from nicely formatted checkboxes and tabular layouts that permit easy reading or make selections easier to mark.

Thus, word processing or desktop publishing software enters the project, and some individuals will begin thinking "Wouldn't it be nice if we could just enter the data for all of the responses back into the forms as a text database, and then somehow print out the results of the survey]" Then, someone will probably respond, "Yes, but word processors don't create good databases. Let's use database software]" At this point the debate begins among the team members on which vendor has the best software to use for the project to create a "survey database." All too often, pragmatism sets in as everyone realizes that most database reporting will require someone to do some computer programming, and in the end the results of the survey will be a set of huge word processing files that are used for simple listings of the contents of the "database." This word processor approach may suffice for many small surveys, however, true text-oriented database software is probably better in most cases. To make a word processor work as a database software will require some sharp skills with primitive text string search capabilities, creating databases built as rows and columns in tables, or even more complex mail merge files. This level of skill requirement seems to indicate that selecting the right software for the job is very important.

Actual database software as Microsoft's Access, Borland's dBase, or Ask Sam Systems' Ask Sam are all much more useful data management environments than a word processor could ever be. These products are specifically designed for data storage, retrieval, and reporting. However, use of each product requires learning the computer interface of the software, creating database structures to hold the data, and mastering each program's individual strengths at information searching and reporting. The degree of difficulty of learning to use these systems depends to some extent on one's previous experience with database software. Those individuals that are entirely new to the use of database systems had better set aside some time for training and hands-on learning. It is true that most contemporary database software has reasonably "user-friendly" interfaces, but most novices should not expect to be entering and reporting data without having had adequate time for learning all of the features needed to feel comfortable using the software. Selecting the right general database software should be done by first consulting some of the numerous computing reviews that are available on this class of software.(1,2,3,4,5)

At this point in a survey project, one may realize that the paper survey form to be mailed out could also be mailed out as a computer file on a disk for the survey respondent to fill in on their own computer. This is an excellent idea in that any disk-based data returned from a respondent could be loaded directly into the survey database, thus precluding the need for considerable amounts of data entry by the survey collectors. The problem with this approach is that the survey generator is now taking on an additional set of duties. Assuming that the survey form to be filled out and the database will fit on one floppy disk, a data entry screen that a survey respondent can use to fill in the data must still be programmed. The disk and database must also be compatible with the computer system that the respondents are running (such as Windows or Macintosh). If the survey form and database will not fit on one floppy disk, an installation program is in order, requiring even more technical expertise of the survey distributors. Of course, it should also be verified that all diskettes for distribution are free of computer viruses before mailout]

For a well defined survey audience that is comfortable using computers to fill in data on a computer screen, an automated survey instrument that is distributed on a computer disk will work very well. It will probably still be necessary to distribute the paper survey form in most cases, in order to give the respondent an alternative way to fill out the form in case their computer system will not accept or work with the database on the diskette. Any distributed automated surveys on diskette will need to be accompanied by help instructions and will work best if they have context sensitive help instructions embedded in the survey program.

SURVEY TECHNIQUES

As with all technology supported activities, one should occasionally reflect on the business side of the processes that will make an undertaking successful. With surveys, it is very easy to forget that there is an immense body of professional thought and literature regarding the statistical analysis of data. It is desirable to assure that the results or conclusions derived from the survey are valid for the data that was gathered and the approach that was used in the actual data gathering process. In fact, it is all too often the case that erroneous conclusions are drawn from limited survey samples or from survey instruments that are unwittingly slanted toward predisposed conclusions. A very common problem is that a survey may be performed by a "computer person," because this person is best trained to create the form and enter the data into a database. Due to lack of experience and training, this person may forget to ask some very important questions about demographics such as age, sex, cultural origin, or occupation that can give considerable insight into the respondent's answers to questions.(6) They may also pick their favorite database package to do the data entry, simply because they will be able to get the work done faster.

Many of the analytical results of a survey are very dependent on the methods used to perform the survey. If one is to determine questionnaire quality with such techniques as factor analysis or multiple regression analysis, then some sophisticated statistical analysis software may be required. In addition, performing even common calculations such as taking an average, determining a standard deviation, or generating cross tabulations are not always easily performed with standard database software. Most textoriented data managers cannot perform any but the simplest of numerical calculations. It may be valuable to consult with a statistician or some formal reference sources before initiating a survey that will attempt to draw complex conclusions from data gathering efforts.(7,8)

SPECIALIZED SURVEY SOFTWARE

Obviously there are some advantages to selecting software that will support all of the steps possible in conducting a survey. Steps five (perform data analysis) and six (report the results of the survey) are the special tasks that some software is dedicated to performing well. This software can perform complex analysis, store the data in industry standard database formats, aid in the creation of the survey instrument, and generate both tabular data reports and graphical output. Since this software is already set up to perform these activities, it will require correspondingly less programming from the user, thus allowing the user to concentrate on the overall data gathering effort, survey distribution, and the survey results.

A traditional package for statistical analysis of data that is used frequently is SPSS Inc.'s SPSS for Microsoft Windows (also available in a DOS version). This software contains numerous statistical modules for advanced data analysis. It is a well-known and full featured product for about $695.(9,10,11) The capabilities of this product may exceed the needs of many individuals who are not expecting to perform complex analysis.

Another product that is specifically targeted to performing surveys is Raosoft Inc.'s Survey. This software package attempts to aid in all of the previously mentioned seven steps used in conducting and reporting data gathered from surveys. This software helps one create the survey, distribute the form, analyze the responses, and print out the results. It comes with questionnaire-type templates for use in surveying an audience or that can be used to develop other survey forms. Standard statistics such as sums, averages, standard deviations, and cross tabulations can be created and data can be displayed in text format or as bar graphs or pie charts. Data is stored in a standard .DBF file format, so that the data can be used by other programs. A mail merge feature enables creating custom letters to survey respondents. At $495, this software can save a lot of the computer programming that would be required to create these features for other database software. The product has received numerous reviews and seems to work well in a variety of environments.(12,13,14)

PERSONAL CHOICES

There are many factors to consider in selecting software for performing surveys. Small surveys can easily be accommodated with a word processor and some forms development skills. A study of a large number of individuals or items, however, may indicate that a database management system is in order so that the data can be sorted and printed out easily once it has been received and entered into an electronic file. If the study will allow extensive comments by the respondents, a text-oriented data manager will allow searching large text blocks and printing out the search results. However, if complex analysis of the data is required, a statistical analysis package is probably in order. As is so often the case, one should define the data processing I requirements well ahead of making a selection decision on the software required to support a project. It should be some small consolation to realize that there is definitely a software package out there somewhere that will do the job]

REFERENCES

1. "Microcomputer DBMSs and application development systems," DBMS 7(6), June 15, 1994, p. 42-47.

2. Stephen W. Plain, "Best database; Paradox for Windows 5.0," Computer Shopper, 15(1), January 1995, p. 180.

3. John Gleidman, "dBase gets a new lease on life," Computer Shopper, 14(12), December 1994, p 370.

4. Christopher O'Malley, "Freeing data from structural limits," Computer Shopper, 15(1), January 1995, p. 371.

5. Celeste Robinson, "Microsoft Access 2.0 adds magic with more Wizards and OLE 2.0," PC World, 12(6), June 1994, p. 86.

6. Gordon Campbell, "Survey Mania," Computig Canada, 20(17), August 17, 1994, p. 11.

7. Earl R. Babbie, Survey Research Methods, Wadsworth Publishing Company, Belmont, CA, 1990.

8. Paul J. Lavrakas, Telephone Survey Methods: Sampling, Selection, and Supervision, Sage Publications, Newbury Park, CA, 1993.

9. Charles Seiter, "SPSS commits to Macintosh: complete product line," Macworld, 12(1), January 1995, p. 35.

10. Alan J. Fridlund, "Latest SPSS for Windows still holds the statistical lead," InfoWorld, 16(43) October 24, 1994, p. 118.

11. Peter Coffee, "SPSS 6.1 for Windows strikes a good balance," PC Week, 11(36) September 12, 1994, p. 81-2.

12. Robert Gryphon, "Manage your survey results," Data Based Advisor, 12(8), August 1994, p. 16.

13. Florence Olsen, "AF re-launches survey, using PCs and BBS," Government Computer News, 13(18), August 15, 1994, p. 46.

14. Bob Gallagher, "Raosoft's Survey reins in statistics," PC Week, 11(5), February 7, 1994, p. 77.

Copyright Association of Records Managers and Administrators Inc. Jan 1996
Provided by ProQuest Information and Learning Company. All rights Reserved