Freiburg Personality Inventory FPI-R. Primary data from the standardization sample 1999.

Research Questions/Hypotheses:

The standardization of 1982 is satisfactorily reproducible in the new 1999 survey, the resulting differences having only a small effect size. The scales of the FPI-R are satisfactory reproducible, with the structure and test statistical values aligning on the most important issues.

Research Design:

Normalized Test Procedure; single measurement

Measurement Instruments/Apparatus:

TheMultistrategic design
The scale development of the FPI-R is both hypothetically deductive, with the necessary range of personality traits essential specific research and clinical areas, and empirically inductive, utilizing both factor and cluster analysis. The approach is trait oriented, and this personality inventory should serve as a medium bandwidth description system for various differential-psychological assessment tasks.
Factor analysis as well as cluster analysis, performed using Ward’s method, was applied to the present dataset (1999). The Rasch scaling method was not used again as these models demand very restrictive conditions. These conditions lead to sometimes undesirable – even negative – consequences for the multifaceted personality characteristics which have not been sufficiently clarified. Even the newer item response models are not regarded as the “method of choice” for the construction of personality traits. See the detailed discussion in the test manual of the 8th Edition (2010).
The entire data (3,740) was used for the construction of the FPI. The analyses were performed with the relevant programs from the SAS (SAS Institute Inc.) and SPSS, the factor analysis with the principal components method and varimax rotation, and cluster analysis via Ward HGROUP (PROC CLUS of SAS, dissimilarity 1-r). Generally, no missing data were replaced, but, depending on the type of analysis, a reduced number of subjects (pairwise exclusion) was used. Variations between results may be due to differences in either the methods, of each algorithm (PROC instructions in SAS or SPSS), or by differences in the treatment of missing-data.

Results of the reconstruction
Comparison of the 2 large representative surveys from 1982 and 1999 showed that the structure of the FPI-R as well as the test-methodological statistics, reliability coefficients, and even the norms are highly reproducible. The grouping of items based on the new factor and cluster analyses corresponds largely to the previous item scale key. Individual differences are discussed in the test manual. The consistency coefficient (Cronbach’s alpha), based on the new normative sample are between 0.73 and 0.83. These coefficients are satisfactory for the 12 (or rather 14) item scales. While higher coefficients would indicate a greater homogeneity, they would show a redundancy of item content. Stability coefficients are from (nonrepresentative) studies.

Standardization
The standardization could be controlled using the representative sample for West Germany in 1982 and 1999 (partial sample). In addition, trends in the mean values of the FPI scales are an interesting find that needs to be cautiously interpreted due to a lack of a real cohort study. Highly significant changes (p <0.001) were identified in 5 FPI-R scales from 1982 to 1999: an increase in the test scores in the life satisfaction, aggressiveness, and openness scales and a decrease in both the social orientation and physical discomfort scales. The comparison of means is controlled by the factors of gender, age, education, and income status (matched-pairs method). The effect sizes of these medium changes are negligible (d = 0.20).

East-West differences
When comparing the 2 FPI-R sets from 1982 and 1999, distinct differences can be seen in several sociodemographic characteristics. This is mainly explained by the expansion of the population following reunification. The east-west differences in the FPI test scores point to a small effect size. Therefore, item and factor analyses were conducted using entire sample. A reduction of the sample to achieve a proportional composition based on each state’s population was foregone.

Structure of the questionnaire
12 personality traits are captured via the 138 items of this questionnaire: life satisfaction, social awareness, performance orientation, self-consciousness, excitability, aggression, stress, physical discomfort, health concerns, and openness as well as the 2 secondary factors (Eysenck) of extraversion and emotionality (neuroticism). The questionnaire items are declarative sentences (“right”, “wrong”). Based on factor and item analyses, 5 political attitude scales were heuristically formed from the 32 attitude items: conservative, liberal, green, proglobalization, and women’s rights.

Data Collection Method:

Data collection in the presence of an experimenter

• Individual Administration
• Paper and Pencil

Population:

Population of the Federal Republic of Germany (old and new states) aged 16 years or older Germans in a reunified Germany

Survey Time Period:

Sample:

Quota Sample

Gender Distribution:

53,4% female subjects (n=1997)
46,6% male subjects (n=1743)

Age Distribution: 16 years or older

Spatial Coverage (Country/Region/City): Germany

Subject Recruitment:

For this study approximately 500 survey assignments, made up of 5 interviews each, were distributed to states and local government districts that were divided according to 6 size classes. Within these regional districts the interviewer was instructed to combine a specific gender with an age bracket as well as a specific employment status in combination with gender and the varying professional circles.
In order to sufficiently analyze the new federal states, these were treated with more consideration than their actual share of the total population would normally mandate. If population-representative data for the whole of Germany is desired, this disproportional consideration must be reversed to the actual population share of 20% as opposed to the 44% used here.

Sample Size:

3740 individuals

Return/Drop Out:

Of the total of 4097 people who participated in the multiple-topic survey from September to November 1999, 3805 (93%) completed the questionnaire. 65 of these questionnaires (1.7%) were excluded: 17 due to inconsistencies between the FPI questionnaire and the IfD record concerning gender and age and 48 due to many missing data in the FPI. In concordance with the recommendations given in the manual, a questionnaire with more than 7 nonresponses (5% of the 138 items) was to be excluded from evaluation. The sample therefore included 3740 persons. The interviewer noted whether the FPI was completed (63.9%) directly or completed later and then either retrieved (28.9%) or returned by mail (2.1%, 5.1% without notice).