SPSS File Guidelines

1) We require a SAV file. 

2) No personal information permitting identification of respondents should be transmitted to Telmar. All data such as name, full street address, telephone number, email address must be removed before delivery. 

3) If the data for a particular variable is not to be used in eTelmar then that variable should be removed. 

4) We do not use alpha text so all variables should be of numeric type. Open-ended or String questions (if required) should be re-coded into a numeric format. 

5) The Variables should always be the minimum necessary Width & Decimals and not the SPSS default of 8 & 2 (F8.2). For example, Gender would be Width 1, Decimals 0 (F1.0). Sometimes it is necessary to have more digits, e.g. Actual Age would normally be Width 2, Decimals 0 (F2.0). 

6) The number of decimal places should be set to zero except for the weights or other such appropriate answers requiring decimal precision. 

7) Variable widths: 

a) Integer variables: maximum width is 10 (F10.0). 

b) Non-integer variables: Maximum overall width is 18 including decimal point (F18.x). 

i) maximum width of the integer part is 10, regardless of the overall length (i.e. F18.7, F17.6,…,F12.1). 

ii) maximum width of the decimal part is 16, only possible if the integer part is 0 (F18.16). 

8) There should always be text in the Variable Label. 

9) The full question text should be entered in the Variable Label. For example, if the question is “Which of these statements do you agree with on a scale of 1 to 5”, the statement text should also be included.

10) If question text includes a secondary question then the label should show the primary question first, with a consistent delimiter joining them, i.e. “::” 

11) The full answer text should be used as the Value Label for each value present in that Variable, unless the answer is an actual number, for example 21 means Actual Age 21, in which case there will be no value labels and SPSS variable view should show “None” 

12) There should be NO negative values used in the data or Value Labels. We only process positive values therefore such values as -99 or -1 need to be recoded with an appropriate value. 

13) Missing answers should ideally be coded as 0 with a value label of “No answer”, if 0 is not already in use as a valid answer value. If 0 is in use, e.g. “Number of foreign trips last year”, then missing values (i.e. non-answers) should be left blank. 

14) Due to analysis techniques in our system, we require a minimum of 256 total respondents (sample size). Where the sample size is lower than 256 respondents, the total respondent count will be increased to 256 in crosstab. 

15) Dataloader currently has an upload size limit of 2GB (browser limitation)