SPSS notes
Week 1
Calculate scale scores
Transform > compute variable > fill in value name > numeric expression: MEAN.7 (n=
number of items a participants should at least have answered -> in this case little over half) >
(lokfa1 to lokfa7, lokfa9 to lokfa11)
Create a report
Analyze > descriptive statistics > descriptives > variables: mean scale scores previously
computed > click relevant descriptives (mean, SD, min., max.) > report values
Graph of distribution
Graphs > legacy dialogue > relevant graph (histogram/boxplot) > variable: mean scale score
(fill in different mean scale scores separately -> so two or more graphs!)
Creating norms of the scales
1. Analyze > descriptive statistics > descriptives > unclick all descriptives > variable:
mean scale scores > save as standardised values
2. Analyze > descriptive statistics > descriptives > variable: newly computed Z-scores >
options: mean, SD, min., max.
3. Transform > compute variable > fill in value name > numeric expression: (Z-score *
10) + 50 > do this for both Z-scores variables
Percentile scores
Transform > rank cases > rank types > unclick rank > check fractional rank as % > ties >
unclick mean > click high > variable: mean scale scores > variable view > decimals for
fractional rank variables 0
Create norm table
Analyze > reports > case summaries > variables: Z-score, T-score, %-score (do this per scale
score!) > grouping variable: mean scale score > unclick display cases > statistics > drag mean
to the right box > drag number of cases back to the left box > check whether total T-score =
50, and Z-score = 0 > repeat all previous steps for other scale score(s)
Week 2
Split-half method
1. Split items subscale in two groups > analyze > descriptives > mean > standard
deviation (do this for all relevant subscales!)
2. Analyze > scale > items: all items group 1 and 2 > statistics > descriptives for: scale >
ok > model: split-half > ok
3. Do the same for all relevant subscales
4. Look at ‘equal length’ for value interpretation
, Calculate reliability for a scale
Analyze > scale > reliability analysis > items: items of relevant scale -> look at Chronbach’s
Alpha > repeat for relevant scales!
Which items contribute most to reliability of a scale?
Look in table > look at ‘Cronbach’s alpha if item deleted’ > which value is the lowest? Or
most apparent?
Which items impair reliability?
Which item is excluded? > does overall reliability cross the found alpha when we remove an
item? > it gets better when removed
!! Look in the column alpha if item deleted for the item with the highest value. If this value is
higher than the alpha for the total scale, you continue to the next assignment, if there is no
such value, you are done
Delete items that impair reliability
Analyze > scale > reliability analysis > add all items of relevant scale EXCEPT for impairing
item > descriptives for: scale if item deleted > look at change in general C. Alpha
Test lengthening or shortening
Find value:
Use the formula -> Rxx-original = before-found Cronbach’s
Alpha > repeat for all relevant scales!
Conclusion:
Look at change of original/revised > which difference (increase) is the most? > this scale
benefits the most from change in length of a test
Week 3
Calculate total score
- Analyze > report > case summary > uncheck: display cases > all variables in variable
box > look at percentage excluded > 0? = no missing values
- Transform > compute variable > SUM(variable 1 to variable [last variable]) > repeat
for all scale scores
Determine correlations between measures
- Analyze > correlations > bivariate > look at order of scales > add all to the right box >
ok
- Double click table > pivot > pivoting trays > drag ‘statistics’ to the left upper box >
select all values > cell properties > format values > change decimals to 2 > ok
- Colour in the table all the types of correlations values (monotrait-monomethod, etc.)
Determine reliability of the measures (monomethod-monotrait)
- Analyze > scale > reliability analysis > put in all items of first scale > ok > repeat for all
scale scores > use Chronbach’s Alpha > fill them in in table
Week 1
Calculate scale scores
Transform > compute variable > fill in value name > numeric expression: MEAN.7 (n=
number of items a participants should at least have answered -> in this case little over half) >
(lokfa1 to lokfa7, lokfa9 to lokfa11)
Create a report
Analyze > descriptive statistics > descriptives > variables: mean scale scores previously
computed > click relevant descriptives (mean, SD, min., max.) > report values
Graph of distribution
Graphs > legacy dialogue > relevant graph (histogram/boxplot) > variable: mean scale score
(fill in different mean scale scores separately -> so two or more graphs!)
Creating norms of the scales
1. Analyze > descriptive statistics > descriptives > unclick all descriptives > variable:
mean scale scores > save as standardised values
2. Analyze > descriptive statistics > descriptives > variable: newly computed Z-scores >
options: mean, SD, min., max.
3. Transform > compute variable > fill in value name > numeric expression: (Z-score *
10) + 50 > do this for both Z-scores variables
Percentile scores
Transform > rank cases > rank types > unclick rank > check fractional rank as % > ties >
unclick mean > click high > variable: mean scale scores > variable view > decimals for
fractional rank variables 0
Create norm table
Analyze > reports > case summaries > variables: Z-score, T-score, %-score (do this per scale
score!) > grouping variable: mean scale score > unclick display cases > statistics > drag mean
to the right box > drag number of cases back to the left box > check whether total T-score =
50, and Z-score = 0 > repeat all previous steps for other scale score(s)
Week 2
Split-half method
1. Split items subscale in two groups > analyze > descriptives > mean > standard
deviation (do this for all relevant subscales!)
2. Analyze > scale > items: all items group 1 and 2 > statistics > descriptives for: scale >
ok > model: split-half > ok
3. Do the same for all relevant subscales
4. Look at ‘equal length’ for value interpretation
, Calculate reliability for a scale
Analyze > scale > reliability analysis > items: items of relevant scale -> look at Chronbach’s
Alpha > repeat for relevant scales!
Which items contribute most to reliability of a scale?
Look in table > look at ‘Cronbach’s alpha if item deleted’ > which value is the lowest? Or
most apparent?
Which items impair reliability?
Which item is excluded? > does overall reliability cross the found alpha when we remove an
item? > it gets better when removed
!! Look in the column alpha if item deleted for the item with the highest value. If this value is
higher than the alpha for the total scale, you continue to the next assignment, if there is no
such value, you are done
Delete items that impair reliability
Analyze > scale > reliability analysis > add all items of relevant scale EXCEPT for impairing
item > descriptives for: scale if item deleted > look at change in general C. Alpha
Test lengthening or shortening
Find value:
Use the formula -> Rxx-original = before-found Cronbach’s
Alpha > repeat for all relevant scales!
Conclusion:
Look at change of original/revised > which difference (increase) is the most? > this scale
benefits the most from change in length of a test
Week 3
Calculate total score
- Analyze > report > case summary > uncheck: display cases > all variables in variable
box > look at percentage excluded > 0? = no missing values
- Transform > compute variable > SUM(variable 1 to variable [last variable]) > repeat
for all scale scores
Determine correlations between measures
- Analyze > correlations > bivariate > look at order of scales > add all to the right box >
ok
- Double click table > pivot > pivoting trays > drag ‘statistics’ to the left upper box >
select all values > cell properties > format values > change decimals to 2 > ok
- Colour in the table all the types of correlations values (monotrait-monomethod, etc.)
Determine reliability of the measures (monomethod-monotrait)
- Analyze > scale > reliability analysis > put in all items of first scale > ok > repeat for all
scale scores > use Chronbach’s Alpha > fill them in in table
