Abstract
The results found in the regression models in the previous chapter confound the idea of a simple mediation effect of the model component ‘Psychosocial Symptoms and Consequences’ that would explain the discrepancy between the Self-Assessed Measure (economic strain: difficulties in making ends meet) and the Perceived Measures (economic stress: worry about not having enough money). According to the results of our research so far it seems indeed likely that the effects of a diminished sense of self and role strains are not linear across levels of economic resources. In such a scenario, the relationship between the Self-Assessed and the Perceived Measure is no longer characterized by mechanisms of mediation but by a more complex type of interaction effect referred to as moderation. We will test this moderation hypothesis by estimating a path model based on our theoretical model that allows the path estimates to vary by categories of wealth. The question of interest is which paths are significantly different for different levels of wealth and which paths can be treated as equal across wealth categories.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
For an explanation of the concepts of moderation and mediation, see Footnote 122, p.158
- 2.
The Stata command for this SEM option is > method(mlmv). This option assumes normality and missingness at random. It was our preferred method for the models based on psychosocial variables because these variables contained a considerable number of missing values and the default method (ml) would have excluded 96 observations. However, because some variables do not follow a normal distribution, the validity of the estimated coefficients was crosschecked by estimating the same model with an estimation method that does not assume normality (stata command is > method(ml) vce(robust)), which resulted in the exact same estimates.
- 3.
For this model, the method (ml) was used because some of the post estimation commands after group comparisons are not compatible with the method (mlmv). Thus, the sample size was reduced by 96 observations with missing values to NÂ =Â 1535.
- 4.
This being said, we have to remember that earlier in this volume we found that neuroticism, the personality trait that is more prone to worrying, was not associated with the Perceived Measure in any statistically significant way.
- 5.
The Berlin Aging Study found a prevalence of anxiety of 4.5% among the elderly community (aged 70 and older) (Schaub and Linden 2000).
- 6.
The stata command is > estat ginvariant
- 7.
For this model, the method (ml) was used because some of the post estimation commands after group comparisons are not compatible with the method (mlmv). Thus, the sample size was reduced by 96 missing observations to NÂ =Â 1535.
- 8.
The stata option utilized is > vce(bootstrap, reps(1000) seed(111)). There was no difference in the estimated coefficients.
Reference
Schaub, R. T., & Linden, M. (2000). Anxiety and anxiety disorders in the old and very old – Results from the Berlin aging study (BASE). Comprehensive Psychiatry, 41(2, Suppl 1), 48–54.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Henke, J. (2020). A Path Model of the Relationship Between the Self-Assessed and the Perceived Measure: Group Comparison by Wealth. In: Revisiting Economic Vulnerability in Old Age. Life Course Research and Social Policies, vol 11. Springer, Cham. https://doi.org/10.1007/978-3-030-36323-9_27
Download citation
DOI: https://doi.org/10.1007/978-3-030-36323-9_27
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-36322-2
Online ISBN: 978-3-030-36323-9
eBook Packages: Economics and FinanceEconomics and Finance (R0)