Development of the Multidimensional Mood State Questionnaire (MDBF). Primary data.

Research Questions/Hypotheses:

Research Design:

Standardized Survey Instrument; repeated measurements

Measurement Instruments/Apparatus:

The construction of the MDBF is based on Latent State Trait Theory (LST-Theory; e.g. Steyer, Ferring & Schmitt, 1992), a generalization of classical test theory that explicitly considers situative effects. In LST-Theory, the manifest test value is decomposed into a “true” state value (state) and a measurement error value. The true state value of a person is the sum of a latent personal value (trait) and a value specific to measurement occasion, dependant on the situation and the interaction between the situation and person. Besides reliability, consistency and variability (or measurement occasion specifity) are defined as further important parameters. Consistency is a measure for the degree to which interindividual differences in test value variables are induced by true differences between people. Measurement occasion specifity is a measure for the degree to which true interindividual differences in test value variables are determined by situation specific and interactional factors, making it a measure for the “change sensitivity” of a test. The consistency and specifity coefficients are proportions of variance, whose sum equals the reliability coefficient. The metrological orientation on which the MDBF is based, takes into account that emotional states are strongly situation-dependant.
Item selection is described in detail by Steyer, Schwenkmezger, Eid and Notz (1991; also see Steyer et al., 1997). An item pool was comprised of the items from 12 German-speaking mood state scales. Every item was evaluated by four graduated psychologists and four psychology students using seven exclusion criteria (e.g. "Does the item solely describe observable behavior?"; "Does the item describe a physically localizable state?"). The item selection criteria were derived from the test authors’ mood state concept and served the purpose of excluding items which are difficult to understand or semantically unsuitable. This evaluation was the basis for the selection of a total of 85 items, which were used in a pilot study with three different response scales (visual analogue scale, dichotomous rating scale, seven-point rating scale) and presented to 544 students at the University of Trier. Thereby, the dichotomous rating scale resulted in the least reliable test values, whereas the other rating scales did not differ significantly from one another. As a result of these empirical findings and in view of analyses with models for categorical variables, the authors decided on a five-point rating scale. Using factor analytic, item statistic and theoretical selection criteria, a preliminary mood state scale with 58 items, which represent the three following mood state areas, was compiled.
This preliminary mood state scale was presented to 503 subjects (aged 17 to 77 years) at four points of measurement (t1-t4), each at an interval of approx. three weeks. An initial version with 32 items (Steyer, Schwenkmezger, Notz & Eid, 1994), comprised of the scales “elevated-depressed mood” (16 items), “awake-tired” (8 items) and “calm-nervous” (8 items) was developed by comparing the factor analytic results to the four points of measurement, item statistic criteria and further content-related criteria. For the final test version (Steyer et al., 1997), the scale “elevated-depressed mood” was renamed “good-bad mood” and shortened by half. In addition, one item from the scale “calm-nervous” was exchanged.
The final version of the MDBF is comprised of 24 adjectives with a five-point rating scale, which correspond to the following three dimensions (conceptualized bipolarly):
(1) Good-bad mood (four items characterize the pole "good mood", four items characterize the pole "bad mood");
(2) Awake-tired (four items characterize the awake pole, four items characterize the tired pole);
(3) Calm-Nervous (four items characterize the calm pole, four items characterize the nervous pole).
Every item is rated on a five-point intensity scale. The first category is labelled “not at all”, the fifth category “very”. The other categories are not labelled verbally. The categories are assigned the numbers one to five.
Besides the long version comprised of 24 items, two parallel short forms were composed, each comprised of 12 items.
For analyses the items from the "negative" mood pole (bad mood, tired, nervous) must be recoded. For every subscale, the values of the corresponding items are added up, allowing the values can vary between 8 and 40 per scale.

Data Collection Method:

Data collection in the presence of an experimenter

• Group Administration
• Paper and Pencil

Data collection in the absence of an experimenter

• controlled snowball system, where hired staff recruit subjects under certain stipulations and the subjects fill out the questionnaires on their own

Population:

Adolescents with chGerman-speaking adolescents and adults

Survey Time Period:

The questionnaire was presented four times in a three-week interval:
t1: 08.04.1991-27.05.1991
t2: 15.05.1991-16.06.1991
t3: 26.05.1991-23.07.1991
t4: 14.06.1991-16.08.1991

Sample:

The sample is a mixture of the snowball method and a survey of subjects recruited in lecture halls at the University of Trier.

Gender distribution: 

58,1% female subjects (n=292)

41,9% male subjects (n=211)                                                                                                                                                                                                                                                                                                         

Age Distribution: 17 to 78 years

Spatial Coverage (Country/Region/City): Germany/Rhineland-Palatinate/Trier

Subject Recruitment:

For the group study, students as well as citizens of the city of Trier were recruited. As it was expected that student participation would be especially high as a result of the short access route and compensation, the number of student participants was limited to a maximum of 200 people. The number of citizens that signed up for participation per telephone as a result of two newspaper advertisements was not limited. The subjects who participated at all four points of measurement received a compensation of 40.00 DM and 10.00 DM for travel costs.
As a further method for the study of subjects not part of a student population, a controlled snowball method was used. Hereby students of University of Trier were hired to recruit approx. five to ten subjects from their circle of acquaintances and organize their study participation A maximum of two of these participants were allowed to be students. The hired study staff passed the questionnaires out to their recruited participants on an arranged appointment and later collected the filled-out questionnaires. The filled-out questionnaires were then returned to the project’s scientific staff and the hired study staff received the questionnaires for the next round of surveys. At the end of the study the study staff received 10.00 DM for every recruited subject whose participated at all four points of measurement. The subjects were compensated for their participation with 40.00 DM. Although this method has the advantage of relatively low organizational effort for the survey of non-student subjects, there are several problems associated with it. For example, there is only a remote amount of control during test response, as the subject is not in the presence of the investigator while filling out the questionnaire. To eliminate as many interference factors as possible, study staff were instructed about the objectives of the study in detail. In addition, the number of subjects to be supervised was limited to a maximum of 20 people. Furthermore, the study staff were selected with care. An agreement was concluded with the study staff as well as the supervised subjects, detailing the required services and corresponding compensation. If the agreement was violated, the compensation was not paid out. Additionally, an address list complied by the study staff allowed for random checks during the study.

Sample Size:

503 individuals

Return/Drop Out:

Of the initial sample of 548 subjects, 511 people participated at all points of measurement. At 7 %, the drop out rate may be considered very low. From these 511 datasets, the data from eight people were excluded because of abnormalities (e.g. incomplete data, response tendencies).