Basics dump stuff
- When opening SPSS always open a syntax & save it
- Homogeneity should kind of look like an American football
- Linear regressions should always be linear between x and y. Otherwise the assumptions are
not met.
- Missing(user)-missing: participant did not want to answer: in data shown as: 9(or999).
- System-missing no answer at all: shown as * in data set/frequency.
- Double click on alpha to see exact value
- Multicollinearity very high correlation in independ variables
- VIF: very high multipolarity
o Look at VIF (should not be higher than 10 for two variables)
o If there are two higher look at collearity diagnostics (at least two variance
proportions should be higher than 0.9)
- Always click run after a paste (otherwise select all and run together) [paste: copying it in
syntax}
Analysis that always needs to be done
1. Creating new data file (data preparation):
New datafile with only few variables that you want to keep. Generally used with simple linear
regression.
a. File save as
b. Click “variables”
c. Click drop all
d. Click in “keep boxes” all variables you want to keep
e. Click continue
f. Save with different same (Pick different name so old file is not used)
g. Open the new file with SPSS
h. Click view data to see new tap variables “variables view to see amount”
B (new syntax)
i. File new syntax
j. Always use paste to put in syntax
k. You can use button or ctr R
l. Save the syntax regular
2. Frequencies (see if values are missing) [missing values will be indicated with a *]
a. Analyse
b. Descriptive statistics
c. Frequencies
d. Add all categorical(dummy)/(continues) variables
e. Past and run
- Note: missing values need to be missing-missing or system-missing when creating dummy
variables.
- Note: this gives you what kind of percentage of data would be useable.
, 3. Scatterplot (simple scatter) [does not work with categorical variables]
(note: size of dot = more people)
a. Graphs
b. Legacy dialogs
c. Scatter/dot
d. Simple scatter
e. Put in X and Y
f. Paste
g. Run syntax
4. Histogram
Dummy variables (categorical 0/1) [0 = reference category / 1-i = other]
1. Look at “values” in the data view tab to see how they are coded (in variable view)
2. Click transform
3. Click recode into different variables
a. Put in gender/variable
i. Name: female/whatever
ii. Label: gender/same as before
iii. Click change
b. Click old and new values
i. Put in old: 2 put in new 1
ii. Put in old: 1 put in new 0
iii. Put in missing-system (if there are missing values you can see in frequencies)
4. Continue, paste, run syntax
- Trying running frequencies with the changed variables to see if it worked.
RECODE gender(1=0) (2=1) (SYSMIS=SYSMIS)into female.
EXECUTE.
Assumptions:
Anova - Correct measurements (categorical
Data preparation: X and continues Y)
- Normal distribution
1. Syntax = file new syntax (if needed) - No outliners
- Approximately equal variance whin
groups
- Homogeneity
- The observations should be
independent (within- and between
- When opening SPSS always open a syntax & save it
- Homogeneity should kind of look like an American football
- Linear regressions should always be linear between x and y. Otherwise the assumptions are
not met.
- Missing(user)-missing: participant did not want to answer: in data shown as: 9(or999).
- System-missing no answer at all: shown as * in data set/frequency.
- Double click on alpha to see exact value
- Multicollinearity very high correlation in independ variables
- VIF: very high multipolarity
o Look at VIF (should not be higher than 10 for two variables)
o If there are two higher look at collearity diagnostics (at least two variance
proportions should be higher than 0.9)
- Always click run after a paste (otherwise select all and run together) [paste: copying it in
syntax}
Analysis that always needs to be done
1. Creating new data file (data preparation):
New datafile with only few variables that you want to keep. Generally used with simple linear
regression.
a. File save as
b. Click “variables”
c. Click drop all
d. Click in “keep boxes” all variables you want to keep
e. Click continue
f. Save with different same (Pick different name so old file is not used)
g. Open the new file with SPSS
h. Click view data to see new tap variables “variables view to see amount”
B (new syntax)
i. File new syntax
j. Always use paste to put in syntax
k. You can use button or ctr R
l. Save the syntax regular
2. Frequencies (see if values are missing) [missing values will be indicated with a *]
a. Analyse
b. Descriptive statistics
c. Frequencies
d. Add all categorical(dummy)/(continues) variables
e. Past and run
- Note: missing values need to be missing-missing or system-missing when creating dummy
variables.
- Note: this gives you what kind of percentage of data would be useable.
, 3. Scatterplot (simple scatter) [does not work with categorical variables]
(note: size of dot = more people)
a. Graphs
b. Legacy dialogs
c. Scatter/dot
d. Simple scatter
e. Put in X and Y
f. Paste
g. Run syntax
4. Histogram
Dummy variables (categorical 0/1) [0 = reference category / 1-i = other]
1. Look at “values” in the data view tab to see how they are coded (in variable view)
2. Click transform
3. Click recode into different variables
a. Put in gender/variable
i. Name: female/whatever
ii. Label: gender/same as before
iii. Click change
b. Click old and new values
i. Put in old: 2 put in new 1
ii. Put in old: 1 put in new 0
iii. Put in missing-system (if there are missing values you can see in frequencies)
4. Continue, paste, run syntax
- Trying running frequencies with the changed variables to see if it worked.
RECODE gender(1=0) (2=1) (SYSMIS=SYSMIS)into female.
EXECUTE.
Assumptions:
Anova - Correct measurements (categorical
Data preparation: X and continues Y)
- Normal distribution
1. Syntax = file new syntax (if needed) - No outliners
- Approximately equal variance whin
groups
- Homogeneity
- The observations should be
independent (within- and between
