Part 3: European Social Survey 2002
Variable names
European Social Survey 2002 (GB only) Data Editor as initialised:
Making it easier to find your way round the file Get a copy of the questionnaire! Modify variable labels to put question number at beginning Adjust columns to necessary basics Change variable names to make them easier to find
You could rewrite all the variable labels from scratch, but for now it was easier to modify them inside the Data Editor
European Social Survey GB only Data Editor after modifying variable labels
Change variable names rename variables (tvtot to pplhlp = a1 to a10) (dscrrce to dscroth = c17_1 to c17_10) (dscrdk to dscrna = c17_dk, c17_ref, c17_nap, c17_na) [The lines in red are for variables used in later examples]
Data Editor with new variable names
Adjust column widths to see more of variable and value labels and mask unneeded columns
Variable labels
An example of awkward labelling European Social Survey 2002
ASK ALL C16 Would you describe yourself as being a member of a group that is discriminated against in this country?Yes1ASK C17 No2 GO TO C18 (Don’t know)8 C17 On what grounds is your group discriminated against? PROBE: ‘What other grounds?’ CODE ALL THAT APPLY Colour or race01 Nationality02 Religion03 Language04 Ethnic group05 Age06 Gender07 Sexuality08 Disability09 Other (WRITE IN)___________________________10 (Don’t know)88
This is an example of a multiple response question (with a preceding filter)
Problems for secondary researcher no indication of data layout mnemonic variable names long variable labels with no question number redundant information at the beginning useful information at the end (and gets lost) binary value labels (0, 1)
Back to the Data Editor as initialised
Here’s what I mean (after scrolling around looking for likely candidates) Could be these: can’t make head or tail of them took a while to find them.
How do I find the right variables? Adjust column widths as before Make Label column even wider to reveal labels in full Scroll down searching for candidates
Data Editor after widening the Label column to reveal variable labels in full This is a sign of lack of experience in SPSS!
How to solve the problem? Step 1: Add question number and response code to beginning of variable labels Step 2: Change variable names Step 3: Get rid of redundant information at beginning of variable labels
Step 1: add question number and response code to beginning of variable label but still with mnemonic variable names
Step 2: change variable names
but there’s far too much redundant information at the beginning of the variable labels and the value labels are binary (0,1) not 0-10,88 as on questionnaire
Step 3: lose redundant info in labels
How do we analyse this question? You could run separate frequency counts for each variable, and then add them all up, but it’s far better to use the SPSS command MULT RESPONSE
Mult response Creates a temporary group variable (which cannot be saved) from several variables In binary mode it uses a single value across all variables in the group and prints tables with variable labels In general mode it uses a range of values across all variables in the group and prints tables with value labels
To run SPSS multiple response in binary mode on the original data mult response groups = discrim 'Reasons for perceived discrimination' (dscrrce to dscrna (1)) /freq discrim.
… which produces: Group DISCRIM Reasons for perceived discrimination (Value tabulated = 1) Pct of Pct of Dichotomy label Name Count Responses Cases Discrimination of respondent's group: co DSCRRCE Discrimination of respondent's group: na DSCRNTN Discrimination of respondent's group: re DSCRRLG Discrimination of respondent's group: la DSCRLNG Discrimination of respondent's group: et DSCRETN Discrimination of respondent's group: ag DSCRAGE Discrimination of respondent's group: ge DSCRGND Discrimination of respondent's group: se DSCRSEX Discrimination of respondent's group: di DSCRDSB Discrimination of respondent's group: ot DSCROTH Discrimination of respondent's group: do DSCRDK Discrimination of respondent's group: re DSCRREF Discrimination of respondent's group: no DSCRNAP Total responses missing cases; 2,052 valid cases
To run SPSS multiple response in binary mode on the modified data mult response groups = discrim 'Reasons for perceived discrimination' (c17_1 to c17_nap (1)) /freq discrim.
…not much clearer! Group DISCRIM Reasons for perceived discrimination (Value tabulated = 1) Pct of Pct of Dichotomy label Name Count Responses Cases C17-1: Discrimination of respondent's gr DSCRRCE C17-2: Discrimination of respondent's gr DSCRNTN C17-3: Discrimination of respondent's gr DSCRRLG C17-4: Discrimination of respondent's gr DSCRLNG C17-5: Discrimination of respondent's gr DSCRETN C17-6: Discrimination of respondent's gr DSCRAGE C17-7: Discrimination of respondent's gr DSCRGND C17-8: Discrimination of respondent's gr DSCRSEX C17-9: Discrimination of respondent's gr DSCRDSB C17-10: Discrimination of respondent's g DSCROTH C17-DK: Discrimination of Respondent's g DSCRDK C17-ref: Discrimination of respondent's DSCRREF C17-nap: Discrimination of respondent's DSCRNAP Total responses missing cases; 2,052 valid cases
…shortening the labels helps, but now the variable name is in twice! Group DISCRIM Reasons for perceived discrimination (Value tabulated = 1) Pct of Pct of Dichotomy label Name Count Responses Cases C17-1: Discrimination: colour or race C17_ C17-2: Discrimination: nationality C17_ C17-3: Discrimination: religion C17_ C17-4: Discrimination: language C17_ C17-5: Discrimination: ethnic group C17_ C17-6: Discrimination: age C17_ C17-7: Discrimination: gender C17_ C17-8: Discrimination: sexuality C17_ C17-9: Discrimination: disability C17_ C17-10: Discrimination: other grounds C17_ C17-DK: Discrimination: don't know C17_DK C17-ref: Discrimination: refusal C17_REF C17-nap: Discrimination: not applicable DSCRNAP Total responses missing cases; 2,052 valid cases
There’s another way of doing it which is much better Temporarily change the codes from binary to sequential Disable missing values Add value labels (first variable only) Use MULT RESPONSE in general mode
As a check on initial values (and not just for this example) you can use list var c17_1 to c17_10 / cases 5.
List C17_1 to C17-10 before recoding (first 5 cases only) C17_1 C17_2 C17_3 C17_4 C17_5 C17_6 C17_7 C17_8 C17_9 C17_ Number of cases read: 5 Number of cases listed: 5
Step 1: Temporarily change values from binary to sequential temp. recode c17_1 to c17_10 (6 thru hi = sysmis) /c17_2(1=2) /c17_3(1=3) /c17_4(1=4) /c17_5(1=5) /c17_6(1=6) /c17_7(1=7) /c17_8(1=8) /c17_9(1=9) /c17_10 (1=10) /c17_dk (1=11) /c17_ref (1=12) /c17_nap (1=13) /c17_na (1=14).
List first 5 cases after recoding List C17_1 to C17-10 after recoding (first 5 cases only) C17_1 C17_2 C17_3 C17_4 C17_5 C17_6 C17_7 C17_8 C17_9 C17_ * Number of cases read: 5 Number of cases listed: 5 (NB: the * = 10: it would print with format F2.0)
Step 2:Disable missing values missing values c17_1 to c17_na ( ).
Step 3: Specify new value labels (1 st variable only) value labels c17_1 (1) 'Colour or race' (2) 'Nationality' (3) 'Religion' (4) 'Language' (5) 'Ethnic group' (6) 'Age' (7) 'Gender' (8) 'Sexuality' (9) 'Disability' (10) 'Other' (11) "Don't know" (12) 'Refusal' (13) 'Not applicable' (14) 'No answer'.
Step 4: Specify group variable and get frequency count mult response groups = discrim 'Q17: Perceived reasons for discrimination' (c17_1 to c17_nap (1,14)) /freq discrim.
Perceived reasons for discrimination This is much clearer (if you can read it!) Group DISCRIM Q17 Perceived reasons for discrimination Pct of Pct of Category label Code Count Responses Cases Colour or race Nationality Religion Language Ethnic group Age Gender Sexuality Disability Other Don't know Refusal Not applicable Total responses missing cases; 2,052 valid cases
To produce the table only for those who actually answered the question, we simply change the mult response command to: mult response groups = discrim 'Q17: Perceived reasons for discrimination' (c17_1 to c17_10 (1,10)) /freq discrim.
Group DISCRIM Q17 Perceived reasons for discrimination Pct of Pct of Category label Code Count Responses Cases Colour or race Nationality Religion Language Ethnic group Age Group DISCRIM Q17 Perceived reasons for discrimination Pct of Pct of Category label Code Count Responses Cases Colour or race Nationality Religion Language Ethnic group Age Gender Sexuality Disability Other Total responses ,773 missing cases; 279 valid cases Group DISCRIM C17 Perceived reasons for discrimination Pct of Pct of Category label Code Count Responses Cases Colour or race Nationality Religion Language Ethnic group Age Gender Sexuality Disability Other Total responses ,773 missing cases; 279 valid cases Perceived reasons for discrimination (valid cases only)
Here endeth the third lesson