Dispositional optimism is a personality trait that reflects the general tendency to expect that good things will happen in the future (Scheier and Carver, 1992). Like concepts such as self-efficacy, resilience, resilient coping, and sense of coherence, optimism can be understood as a resource variable that buffers aversive life events. Optimism is associated with mental and physical health (Schou-Bredal et al., 2019), quality of life (Liu et al., 2021; Marton et al., 2022), education and physical activity (Craig et al., 2023), positive adjustment to stressful life events and coping (Ramírez-Maestre et al., 2019), life satisfaction (Cerezo et al., 2022), spirituality (Ciria-Suarez et al., 2021), low levels of anxiety and depression (Faye-Schjøll and Schou-Bredal, 2019; Menéndez-Aller et al., 2020), low risk of burnout (Tutte-Vallarino et al., 2022), low perceived risk for COVID-19 (Schou-Bredal et al., 2021), and even lower mortality (Liu et al., 2022a). The relevance of optimism has been shown in several clinical areas such as heart diseases (Huffman et al., 2019), cancer (Liu et al., 2022b; Hinz et al., 2023), gynecology (Morán-Sánchez et al., 2021), obstetrics (Giangiordano et al., 2020), and chronic pain (Esteve et al., 2018). Recently, optimism (and pessimism) have been discussed as anticipatory feelings related to the process of anticipation and prediction of future events (Stefanova et al., 2020).

The most frequently used questionnaire for measuring optimism is the Life Orientation Test-Revised (LOT-R; Scheier et al., 1994). It consists of two subscales, optimism and pessimism. Multiple studies have been performed that tested the two-dimensional structure of the questionnaire (Cano-García et al., 2015; Hinz et al., 2017; Huang et al., 2020). Until now, it is a matter of debate whether it is justified to combine the two subscales of the LOT-R to one total score (Cano-García et al., 2015; Hinz et al., 2022b). While the optimism and the pessimism subscales are only weakly negatively correlated (Glaesmer et al., 2012; Hinz et al., 2022b), the correlations of the total score with other related constructs are generally somewhat higher than the correlations of the subscales (Hinz et al., 2022b), which is an argument for using this total score.

Normative values for the LOT-R are available from several countries (Zenger et al., 2013; Hinz et al., 2017; Schou-Bredal et al., 2017). These normative studies also include analyses of age and gender differences in optimism in general population samples. However, the knowledge about long-term stability in optimism and the temporal stability of this construct is limited (Armbruster et al., 2015; Saboonchi et al., 2016). Changes in mental health or in resource variables such as optimism can be considered from two different perspectives: mean score changes from t1 to t2, and test–retest correlations which indicate the individual stability, i.e., the stability of the rank positions of the participants. It is well-known that the temporal stability of personality traits is somewhat higher than the stability of health-related variables (Anusic and Schimmack, 2016; Struijs et al., 2020). However, a thorough investigation of long-term changes in optimism, including age and gender differences in temporal stability, has not been done so far.

As mentioned above, habitual optimism is correlated with multiple other health-related variables. However, it is also relevant to investigate not only cross-sectional but also longitudinal associations: To what degree do changes in optimism correspond with changes in other variables? There are only few studies that investigated such associations between change scores of mental health variables (Kuehner, 2002; Hajek and König, 2016), and such change score correlations have not been applied to optimism yet.

In 2017, the results of a study with the LOT-R were published, based on a large representative community sample of 9,711 persons (Hinz et al., 2017). Six years after this baseline assessment (t1), a follow-up-study (t2) was performed which also included the LOT-R. Based on the data of this longitudinal study, the aims of this article were (a) to investigate changes in habitual optimism over a 6-year period, (b) to analyze gender and age differences in the stability of the LOT-R scores, and (c) to investigate the relationship between changes in the LOT-R score and changes in other health-related variables.

The main objective of this study was to investigate the course and the temporal stability of the LOT-R scores over a 6-year period. The LOT-R mean score increased from 16.65 to 17.09 (d = 0.11). This improvement is in line with a study that detected an improvement of mental health (as measured with the SF-8) during a 5-year period (Hopman et al., 2006), but it contradicts other studies which found decreases in mental health (Hemingway et al., 1997). In this context, it is interesting to note that cross-sectional studies on the distribution of optimism showed a decrease in optimism with increasing age (Hinz et al., 2017), which indicates that age dependencies, when derived from cross-sectional studies, can result in other conclusions than those of longitudinal studies which aggregate over individual changes. Further research is needed to clarify the conditions under which long-term increases or decreases of mental health variables occur.

There was a gender effect in the changes of the LOT-R scores: The improvement of the total sample (diff = 0.44) was mainly due to an improvement of the males (diff = 0.66), while the females showed only a slight increase (diff = 0.24). Regarding age, there was a clear and non-linear effect of age group on the LOT-R change score. While the oldest age group (70 years and above) remained relatively stable in the LOT-R mean score level, all other age groups showed an increase, and the strongest increase was found for the age group 60–69 years, which may be due to the beginning of retirement (Prakash et al., 2022; Henning et al., 2023). A decrease in mental health in the oldest age group (75 years and above) was also found in a general population study with a five-year interval (Hopman et al., 2006). Since the increases or decreases in optimism do not follow a linear age trend, statistics based on linear age dependencies are insufficient tools for describing such changes. Particular attention should be given to people in the age group of 70 years and above.

The temporal stability in terms of test–retest correlations of the LOT-R total score was r = 0.61 for the total sample. There are findings on the temporal stability of the LOT-R in several samples of patients, with coefficients between 0.60 to 0.69, e.g., (Hinz et al., 2022b), however, it is problematic to compare these coefficients with those of our study since the time interval was shorter in the patients’ samples (in most cases 3 months). Furthermore, in groups of patients, the individual fluctuations in optimism and mental health will be higher than in the general population due to the different courses of the diseases. A study with university students found a test–retest correlation of r = 0.74, based on a relatively short time interval of only 5 weeks (Trottier et al., 2008), which is understandably higher than the coefficient obtained in our study. Comparisons of such stability coefficients should always take into account the characteristics of the sample and the time interval between the measurements.

While there were only small differences in temporal stability between males and females in general, there were remarkable age differences. The highest temporal stability was found for the age range 40–59 years, and the most pronounced fluctuations were observed for the oldest group (70 years and above). This means that beginning at the age of about 70 years, the prognoses of the individual well-being and optimism become superimposed by other effects.

For all gender and age groups, the stability coefficients of the LOT-R total score were higher than those of the subscales optimism and pessimism. This also confirms that the combination of optimism and lack of pessimism into a sum score is useful, even if CFAs indicate the preference for a two-dimensional model.

The LOT-R scores were statistically associated with multiple other quality-of-life-related variables. As was to be expected, the association between the LOT-R score and satisfaction with life was relatively strong (r = 0.50 at t1), which is comparable with the coefficient (r = 0.45) obtained in another general population study with a similar instrument for measuring life satisfaction (Glaesmer et al., 2012). The associations of the LOT-R total scores with the Mental Component Score of the SF-8 were higher than those with the Physical Component Score both at t1 and at t2 as well as regarding the change score, which underlines the strong relationship between optimism and the mental component of QoL even in the longitudinal perspective. This result is also consistent with the finding that LOT-R optimism captures a positive current state of mind nearly just as strongly as it captures the expectation of future positive events (Hinz et al., 2022a).

A secondary finding was that the correlations obtained in the t2 examination were generally somewhat higher than those of the t1 assessment (Table 4). Evidently, the t2 assessments were somewhat more consistent than those of t1, which is also reflected in slightly higher Cronbach α coefficients at t2 in comparison to the t1 α coefficient. This higher consistency at t2 may be due to the fact that the participants completed the LOT-R at the study center during their t1 examination, while at t2 they completed it at home, which might have contributed to a more consistent way of answering the questions.

In the context of the longitudinal analysis, it is interesting to relate changes in optimism to changes in the QoL and mental health variables. All correlations of the change scores were in the same direction as the cross-sectional correlations; increases in the LOT-R score were associated with increases in life satisfaction, social support, and all components of QoL as well as with decreases in anxiety and in bodily complaints. All these correlation coefficients were lower than the cross-sectional coefficients, a result which has also been found in other studies using such correlations of change scores (Kuehner, 2002; Hajek and König, 2016). Causal interpretations of the correlations of the change scores are not possible: an increase in optimism may be a consequence of a decrease in bodily complaints, and vice versa. A special feature of change score correlations compared to cross-sectional correlations is that they are not affected by response sets. While the positive cross-sectional correlations between two health variables can be, at least in part, due to certain response sets or acquiescence tendencies (Hibbing et al., 2019), these possible response tendencies cancel each other out when individual change scores are calculated as the differences between individual scores.

Some limitations of this study should be mentioned. The sample of the study participants was not perfectly representative of the general population. First, the participants of the baseline assessment were slightly healthier than those who refused participation at t1 (Enzenbach et al., 2019). Second, the comparison between those who dropped out and those who completed both the t1 and the t2 assessment showed an additional bias toward healthy participants. This is a general problem in longitudinal studies and limits the generalizability of the mean score levels, however, the changes over time which are the main topic of this study are less affected by such bias effects. A further general problem of the LOT-R is that both subscales are only weakly negatively correlated, and that the calculation of a sum score may therefore be criticized. Nevertheless, we believe that the superiority of the sum score over the subscales in terms of higher Cronbach’s α coefficients, higher stability coefficients, and higher correlations with the other QoL scales justifies the use of this sum score.

The LOT was designed as an instrument to measure habitual optimism in terms of a stable personality trait. This raises the question of whether such an instrument is suitable at all for recording changes. Recently, a State Optimism Questionnaire (SOM) was presented (Hoeppner et al., 2022), which was tailored to detect individual changes in optimism. However, we believe that the LOT-R is nevertheless the more suitable method, because this questionnaire enables both the assessment of states and changes (by means of differences) in optimism and thus also the comparison of cross-sectional and longitudinal correlations. In addition, unlike the SOM, there are LOT-R norm values and extensive psychometric tests.

Taken together, the results of this study show the degree to which mean score changes and individual score changes of optimism and pessimism occur in the general population. These data can be used to better interpret health changes in samples of patients. People or patient groups older than 70 years deserve special attention because of the limited temporal stability of their optimism scores.

The datasets presented in this article are not readily available because data cannot be shared publicly because of data protection reasons. Data availability is only possible via a project agreement. Data are available from the LIFE-Adult Study Institutional Data Access / Ethics Committee (contact via https://ldp.life.uni-leipzig.de/) for researchers who meet the criteria for access to confidential data. Requests to access the datasets should be directed to https://ldp.life.uni-leipzig.de.

The studies involving humans were approved by Ethical Review Board of the Leipzig University. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

AH: Validation, Writing – original draft, Writing – review & editing. MF: Formal analysis, Writing – review & editing. HG: Investigation, Writing – review & editing. BB: Visualization, Writing – original draft. YN: Data curation, Methodology, Writing – review & editing. MS: Investigation, Writing – review & editing. JE: Conceptualization, Data curation, Writing – review & editing. AM-T: Resources, Writing – original draft.