Notes
a list of old notes on the earlier analyses:
- Hungary deleted because only method 1 and 2 implemented
- Luxembourgh deleted because only method 1 implemented
- In Italy people were asked all questions??? Some answered both method 2 and 3???
- It might be an idea to also look at proportion of item-missing of method 2 compared to method 3.
- The # people participating in the experiments differs a lot across the countries. Variance of the estimates and power differ.
- The number 66 was not coded missing on the trust questions in some countries.
- Sweden has nonrespondents in the dataset with all values missing.
- High correlations in Greece, Austria, Portugal
- Correlation TRSTLGL-TRSTLGL3 in Norway is low (0.553)
Portugal doctors experiment wrong formulation: "rarely" is missing from the translation so the sentences mean something completely different
Classic MTMM model w/ Mplus unstandardised. Equate across countries and look for differences. For those countries + an extreme other one (not scandinavian) go on w/ thresholds.
start with pol trust? or doctors.
write about portugal.
- Greece: 1 person has score 15 for SPLTADMA
- Ireland, Netherlands, Italy administered SB Self-completion rather than F2F
- IS missing SB questions TESTA28-40
- France did all, inc. SC AND FF.
- The women experiment was not done correctly in Portugal:
The i8-10 questions were asked identically to the main questionnaire rather than
as reversed items as stipulated in the source supplementary questionnaire.
Also the source supplementary shows single questions whereas PT used a battery.
- None of the self-completion questionnaires include a Don't Know option, while some of the F2F versions of the same questions do. France may provide an oppurtunity to investigate the effect this has.