Health locus of control (HLOC) is a psychological concept concerning control beliefs in relation to illness, sickness and health. This concept is based on the general approach of locus of control (LOC) developed within the social learning theory by Rotter [1, 2]. General LOC is of fundamental importance in psychology, relevant in established approaches concerning depression [3, 4] and helplessness . Moreover it is closely connected to the concept of self-efficacy . Due to the assumption that general control beliefs could differ from control beliefs concerning health, the specific construct of HLOC was developed and received increased attention in health research over the last 30 years [7, 8]. HLOC research results have been important in understanding health related behaviours, outcomes and care . One main interest of this field of research concerns the compliance of patients in medical care in order to understand patients' adherence to recommended treatments including medication and health related behaviour .
In line with the general construct of LOC, HLOC research assumed a three-dimensional construct with control beliefs concerning Internality, Externality powerful Others (POs) and Externality Chance (Chance) [7, 8]. HLOC research revealed that patterns of HLOC scores differ for patients with specific diseases [10–12]; furthermore, higher scores on Externality scales seemed to be associated with less education [e.g.[13, 14]. The Multidimensional Health Locus of Control scales (MHLC) , parallel Forms A (MHLC-A) and B (MHLC-B), were primarily used in order to assess the three HLOC dimensions . MHLC-A and -B were equivalent, developers reported corresponding correlations for the scales of both forms (for Internality scales r = 0.801, for POs Scales r = 0.761 and for Chance scales r = 0.734) .
The MHLC scales were applied to different languages [e.g.[16, 17] and cross-cultural differences in HLOC were investigated. A study which compared Asian women to British Caucasian women found higher scores for the Asian women on both Externality scales in line with the study's expectations . These results show the Asian cultures stronger beliefs in communal values such as the importance of assisting others as well as the belief in fate as compared to more individualistic western cultures. Surprisingly, the Asian women also revealed stronger Internality compared to western women. This difference was the result of a stronger religiosity of the Asian women indicating culturally different interpretations of the MHLC items: Asian women with a strong belief in 'Allah' had simultaneously strong beliefs in their own actions assuming to help themselves by trusting in 'Allah'. However, the authors argued that the structure of HLOC has not been investigated by factor analysis in a corresponding sample, i.e. the HLOC construct may differ structurally over cultures .
In western cultures, several factor analyses using mostly selected clinical samples confirmed the three-dimensional structure [e.g.[19–21]]; however, a number of studies failed [e.g.[22, 23]. Another western study analysed a mixed clinical sample (N = 588) and detected a four-dimensional HLOC structure developing and validating Form C of the MHLC (MHLC-C) for condition-specific measuring . The authors confirmed original scales Internality and Chance, but had to differentiate POs scale into one scale concerning doctors and the other concerning family and friends. A latter clinical study examined the Italian version of the MHLC-C in a sample of HIV+ patients (N = 478) via methods based on structural equation modeling (SEM) and showed the superiority of the four- over the three-dimensional construct . However, the four-dimensional construct is not implemented in HLOC research and has not been investigated in a non-clinical sample by now.
To our knowledge, the HLOC construct has never been factor analysed utilising a representative general population sample throughout 30 years of research. Diseases and clinical symptoms concerning a great variety of conditions are given in the general population, but the majority should be healthy. An investigation of the HLOC structure in a western general population sample could provide important results for future research and help to predict and understand the compliance of individuals concerning public health care. In addition, such a study could offer an orientation for non-clinical research. Previous factor analyses using non-clinical samples investigated selected student populations [26–28], the staff of a psychiatric hospital  and employees of a university voluntarily participating in a health promotion program . However, a study investigating the general population is lacking.
The aim of the present study was to compare the three- and the four-dimensional construct of HLOC on grounds of a large general population sample representative for a region in Northern Germany by means of ordinal factor analyses based on SEM. We expected to find superiority of the four-dimensional construct and we aimed to confirm our results by cross-validation. Concerning the realisation of our analyses and their presentation we followed guidelines of statistical researchers [31–33]. Our analyses aimed to follow a confirmatory approach comparing theory based models which represented HLOC constructs of differing dimensionality.