Previous Next

ANALYSING survey results is not easy and a computer of some sort is usually pressed into service. Now, with Survey II from Lobos Software, available from Microworkshops, your humble Spectrum or 16K ZX81 can be used to do the job.

Survey II is designed to analyse those surveys with up to six possible answers to each question, and where only one of such answers is valid. Up to 2500 10-question questionaires can be dealt with on the 48K Spectrum.

No allowance is made for storing data on microdrives. The program is designed to drive a ZX printer or a full-sized printer with ZX Lprint III interface. The manual states that by altering the program other interfaces can also be used, although I could not get the Tasman interface to work.

The most horrendous part of any analysis program is entering the data, and Survey II is no exception. When starting on a new survey the number of questionaires, the number of questions, and the number of possible responses to each question are entered. If you try to define more data than the program can cope with you are politely told so, and offered the chance to buy, for a further £10.00 a version which will handle up to 3000 questionaires of 10 questions but with fewer error traps.

The data can then be entered as a string of numbers. A flashing figure at the top of the screen indicates which case number the data is for. If too many, or too few, figures are entered the data is rejected and must be re-entered.

The program checks for data outside the specified range. Unanswered questions are coded with a '9'. The tedium of data entry can be banished to some extent by providing a SAVE option which allows partially entered data to be saved together with the program.

Once the data has been entered it is displayed on the screen and you are given an opportunity to change incorrect data. That can be done at any stage within the program.

Survey II provides two analysis options. The first displays the frequency of different responses to questions; the percentage and cumulative percent are shown also. The results can be displayed as a 3D histogram.

The second shows the frequency of combinations of responses to two questions - crosstabulation, a statistic showing the likelihood of the results occurring by chance can also be obtained.

A full printout of all the data, frequency and crosstabulation charts can be obtained if you have the time. With the sample provided, with 300 cases of six questions, it took over half an hour to printout.

Considering the slowness with which frequencies, for even a medium-sized sample, are calculated the screen display is uninformative. That, and the fact that program and data cannot be saved to microdrive, are my only criticisms. Otherwise Survey II is a solid program which is well protected against accidental crashes.