Evaluation of the factor structure of the Adult Manifest Anxiety Scale--Elderly Version (AMAS-E) in community dwelling older adult New Zealanders.
Anxiety in older adults is a phenomenon that has gained increasing empirical attention. Early investigators in this field asserted that the measurement of anxiety in older adults using the psychometric measures available was problematic due to (1) a lack of construct validity; the field did not have a clear empirical understanding of the features of anxiety in older adults, (2) a lack of psychometric measures designed specifically for older adults--e.g. the inclusion of items of low relevance for older people and a high potential overlap with measuring medical symptoms, and (3) a lack of normative information for existing psychometric measures when used in older adult populations (Stanley & Beck, 2000). The research community responded in two ways, either by developing new measures of anxiety specifically for older adults (e.g the Geriatric Anxiety Scale (Pachana et al., 2007), or by validating existing anxiety measures in older adult populations (e.g. the State-Trait Anxiety Inventory (Spielberger, Gorsuch, & Lushene, 1970). As yet, no 'gold standard' anxiety measure has been identified for use with older-adult clients.
Given that there has been limited research into the nature of anxiety in older adults, the evaluation of the construct validity of measures is problematic as essential content has not been fully explored. Some authors suggest that measures for older adults should: (a) avoid items with worry content (e.g. topics) as they may be less useful in predicting worry severity when compared to worry content (Diefenbach, Stanley, & Beck, 2001); (b) avoid items which overlap with medical conditions (Wolitzky-Taylor, Castriotta, Lenze, Stanley, & Craske, 2010); (c) formats should be age-appropriate for older people with sensory and or cognitive impairments; (d) include developmentally appropriate themes e.g. avoid topics relating to work which may not be relevant to the retired; (e) be sensitive to functional impairment; (f) include sleep disturbances which are common in older adults with generalised anxiety disorder (Wetherell, Le Roux, & Gatz, 2003).
The Adult Manifest Anxiety Scale-Elderly Version (AMAS-E) (Lowe & Reynolds, 2000) is a 44 item measure of chronic manifest anxiety developed for older adults in response to concerns that the nature of anxiety is different in older adults when compared to younger adults. Exploratory factor analysis revealed a three-factor structure: fear of aging --assessing a preoccupation with age-related decline; worry/oversensitivity --reflecting excessive worry; and physiological anxiety--evaluating physical manifestations of anxiety. A 'lie' scale intended to contribute to validity and measure social desirability was added at a later stage, (Lowe & Reynolds, 2006), however the theoretical justification for this was not presented by the authors. The four factors of the AMAS-E all were hypothesized to contribute to a higher order factor of general anxiety/total anxiety (Lowe & Reynolds, 2006). Confirmatory factor analytic (CFA) studies of the AMAS-E have not been published. Psychometric studies of the AMAS-E have shown good reliability for the total anxiety ([alpha] = .90 - .92), worry /oversensitivity (a = .88 - .91), fear of aging ([alpha] = .78 - .85), physiological ([alpha] = .69 - .71), and lie scales ([alpha] = .73 - .79) (Lowe & Reynolds, 2000, 2006; Reynolds, Richmond, & Lowe, 2003).
Convergent validity analysis of the AMAS-E has been undertaken using the State-Trait Anxiety Inventory (STAI) (Spielberger et al., 1970) and the Geriatric Anxiety Scale (GAS) (Segal, June, Payne, Coolidge, & Yochim, 2010). Low to moderate correlations were reported between the anxiety subscales of the AMAS-E and the STAI-S (.24 and .39), and between the AMAS-E and STAI-T (.31 and .65) (Lowe & Reynolds, 2006). Because the STAI had a large amount of variance when measuring anxiety in older adults, it was unclear whether the correlation observed between these measures is due to shared variance unrelated to the construct of anxiety. Moderate correlations have been observed between the GAS total and AMAS-E total (.77), however subscale performance of the AMAS-E varied when correlated with the GAS total (.45 fear of aging, .65 physiological, and .76 worry) (Segal et al., 2010). This suggests more investigation is needed on the performance of the AMAS-E in older adult populations.
Overview of Confirmatory Factor Analysis
Confirmatory factor analysis (CFA) is a form of structural equation modelling that takes a hypothesis testing approach to the analysis of relationships between observed and unobserved variables. CFA is useful when evaluating the performance of psychometric measures as it enables the evaluation of the magnitude and direction of the relationship between items (observed variables), and subscales (factors). It also predicts how well these hypothesised models will perform in the population (Byrne, 2001).
The present study intends to evaluate the psychometric properties including the factor structure of the AMAS-E in a sample of community dwelling older adult New Zealanders.
Ethical approval was obtained through Massey University Human Ethics Committee Northern.
Older adults aged 60 to 80 (M = 68, SD =7.2) were recruited from older adults' community organisations in the North Auckland region, New Zealand. The organisations provided a short verbal presentation slot for the researcher, where the purpose of the research was explained. This was followed by a question and answer session. A second group of participants were recruited through indicating on a previous research study that they would be willing to be contacted for participation in future related research.
Age 60 was chosen to capture the cohort of baby boomers entering older adulthood. Cognitive screening was not practical as part of the study therefore participants who were aged over 80 or residing in rest-homes or hospitals were excluded due to the base rates of significant cognitive impairment in these groups. Of participants who self-selected into the study, 203 participants responded (83% return rate; male n = 73; female n = 117; not specified n = 13) of New Zealand European decent (Pakeha; n = 122); other European (n = 60); Maori (n = 4) and not specified (n =17). 11% of participants reported a psychiatric history, of those 6% reported depression and 6% reported anxiety. The mean anxiety scores on the AMAS-E were in the "normal" range based on the recommended cutoffs (Lowe & Reynolds, 2006). As this study had missing demographic data, an independent samples t-test was conducted to determine if there was a significant difference in scores on the AMAS-E between those who returned demographic information and those who did not. No significant difference was observed between those who had missing demographic information (M = 48, SD = 7.3) and those who did not (M = 46.8, SD = 7.8) t(167) = -1.29, p >.05.
Adult Manifest Anxiety Scale-Elderly version (AMAS-E)
The AMAS-E is a 44 item measure of anxiety in the elderly. Items on the AMAS-E are divided into four subscales: Worry/Oversensitivity (23 items); Physiological anxiety (7 items); Fear of Aging (7 items); and Lie (7 items). Response choices are a dichotomous yes/no format and are designed to assess cognitive, physiological, and behavioural aspects of anxiety. An example of a question is item 22 from the worry scale "I worry a lot of the time". The Lie scale is intended to evaluate concealment of anxiety through social desirability factors which are considered problematic for older adult populations (Reynolds et al., 2003).
Hospital Anxiety and Depression Scale (HADS)
The HADS (Zigmond & Snaith, 1983) is a 14 item self-report measure of anxiety and depression, with items divided equally between both scales. Clients are asked to underline the reply which is closest to the way they are feeling. Each item has a different range of responses, some of which are specifically worded to reflect the item stem. The HADS was chosen as it has evidence of reliability and validity for use with older adults, and has few items that overlap with medical symptoms (Roberts, Fletcher, & Merrick, 2014). As the HADS measures both anxiety and depression it can provide evidence of convergent and discriminant validity for the AMAS-E.
Community organisation members who received a short presentation on the nature of the research, self-selected into the study through collecting set of questionnaires, information sheet, and consent form from the researcher afterwards. A second group of participants received this package of questionnaires following their indication on a partner study that they would be interested in completing further research. Participants also completed the Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983); and a third anxiety measure undergoing development as part of a larger study evaluating assessment measures of anxiety in older adults. The order of measures was randomised to reduce order and fatigue effects.
Participants completed these measures independently at home and returned them by prepaid mail to the researcher. The order of questionnaires was manually changed during collation as per table 2, however participants could complete these in their own time and were not given instructions on completing them in order.
A four factor structure represented by a higher order factor of latent anxiety was created. The complete item set was used and the original item placement on subscales (factors) was retained (Reynolds et al., 2003), see Figure 1.
The same four factors and item placement as AMAS1 were predicted to be correlated, see Figure 2.
A revised correlated model with the lie scale removed.
Item total correlations, internal consistency, and convergent validity were undertaken using SPSS and AMOS version 23.
Structural equation modelling (SEM) requires large sample sizes to generate a level of power necessary for the analysis and ensure accuracy of fit indices and parameter estimates, particularly when the model is complex (MacCallum, Widaman, Zhang, & Hong, 1999). A null hypothesis was specified that the root mean squared error of approximation (RMSEA) was 0.05 or less (i.e. close fit) and an alternative hypothesis that the RMSEA was 0.08 (mediocre fit). The power analysis for the test of close fit of the RMSEA to the population with the present sample of 203 participants and model complexity (898 Df; p <.05) yielded power of >.99% (Preacher & Coffman, 2006).
Missing data within this sample was small and only accounted for 6%. A significant Little's test further suggested that data were missing completely at random (MCAR) and therefore the use of the maximum likelihood method in SEM was appropriate (T. A Brown, 2006). Pairwise exclusion was used in SPSS where available.
Evaluation of Model Fit
The fit statistics chosen for this study were sensitive to model misspecification and provided information pertaining to the fit of a measurement model within the sample and an estimate of the model's fit within the population (Beck, Brown, Steer, & Weissman, 1991). Goodness of fit was assessed using the X2 test of exact fit, the comparative-fit index (CFI), and the Tucker-Lewis index (TLI). Both the CFI and TLI are indicators of model fit derived from a comparison between the hypothesised model and a null model in which all observed variables are uncorrelated. Both indices yield a coefficient with values ranging from 0 to 1, with values greater than .90 suggesting adequate fit, and those exceeding .95 suggest the model is a good fit (Hu & Bentler, 1998). The CFI and TLI are not systematically related to sample size (Fan, Thompson, & Wang, 1999), and the TLI appropriately rewards model parsimony (Cheung & Rensvold, 2002). The root mean square error of approximation (RMSEA) was also used as an indicator of how well the model may hold in the population if the covariance matrix of the population was available. A RMSEA of <.05 is considered a close fit, .05 -.08 indicates reasonable fit, .08 to .10 indicate mediocre fit, and >.1 indicate poor fit (Hu & Bentler, 1998). Unlike the CFI and TLI, the RMSEA is not affected by model complexity (Cheung & Rensvold, 2002) and is minimally affected by sample size (Fan et al., 1999).
The total anxiety scale of the AMAS-E showed good internal consistency (a =.88), however the internal consistency of the subscales was variable (see Table 3.).
The physiological subscale showed low internal consistency ([alpha] = 0.52), whereas the lie ([alpha] = 0.71) and worry/ oversensitivity ([alpha] = 0.88) and the fear of aging ([alpha] = 0.76) subscales showed good levels of internal consistency.
Normality of Data
Evaluation of total scores on the AMAS-E showed data were marginally negatively skewed (-.302) and slightly kurtotic (1.429), and the Kolmogorov-Smirnov test was significant. However, as item level analysis suggested the majority of item responses were normally distributed, and as skew and kurtosis were minimal, the use of parametric statistics can be justified.
Moderate positive correlations were observed between the HADS anxiety subscale and both the AMAS-E total and AMAS-E worry/oversensitivity subscales (.62 and .71; p <.01 respectively; see Table 4.). The fear of ageing subscale and the physiological subscales showed weak relationships with the HADS anxiety total subscale (.42 and .35 respectively; p <.01). A strong positive correlation was observed between the worry/oversensitivity subscale and the AMAS-E total score (.90; p <.01). The lie scale showed little relationship with any subscale the HADS or AMAS-E (-.07 to .05; p >.05).
Confirmatory Factor Analysis
The AMAS1 and AMAS2 models both exhibited similar fit (see Table 5.). Reasonable RMSEA fit indices but low TLI and CFI were observed for both models. A major issue with both models was the low relationship between the Lie scale and the other subscales. All correlations for the Lie scale with other scale were either negative or virtually zero. The latent factor correlations for the AMAS2 ranged between r=-.004 to 0.52. On the basis of these low correlations and the low factor loading in AMAS1, a third model (AMAS3) was specified with the Lie scale being removed. (Table 3). Even with the Lie scale removed the latent factor correlations on AMAS2 between the remaining factors did not change. The removal of the Lie scale did not improve the model fit, but from purely a statistical and theoretical point its removal was warranted. For the AMAS1 and AMAS 2 the average factor loading was [lambda] = 0.51 and for AMAS 3 the average loading was I =0.49.
The internal consistency of AMAS-E total as measured by Cronbach's [alpha] was good, however variability was observed at subscale level. The Worry/ Oversensitivity, Fear of Aging, and Lie subscales showed good internal consistency, however poor internal consistency was observed for the Physiological Anxiety subscale. The homogeneity of some items may have overinflated some of the Cronbach's a on the AMAS-E. The reliability of the AMAS-E in older adult populations is good and demonstrates that true score estimation for three of the four factors is attainable.
The factor structure reported by Lowe and Reynolds (2000, 2006) was not strongly supported in this sample. The specification of AMAS1 and AMAS2 did not provide a clear picture of whether the AMAS-E is best explained by a higher order model (AMAS1) or a fully correlated model (AMAS2). In essence, the presence of moderate to strong factor loadings at the higher order (AMAS1; Table 6) are indicative of the strength of the correlations at the lower order (AMAS2; Table 7). On the basis of the seemingly contradictory fit statistics e.g. the RMSEA suggested acceptable fit, and low CFI and TLI suggested poorer fit, one might reject both of these models. However the divergence in the fit statistics might be explained by the magnitude of the factor loadings in each model. The trialled models found moderate factor loadings. According to Heene, Hilbert, Ziegler and Buhner, (2011) lower fit indices may be expected when there are low factor loadings in a specified model. Thus there can be divergence between the low RMSEA (good fit) and low CFI (poor fit).
Approximately half of the items on the AMAS-E showed weak relationships with their respective factors, or were problematic for other reasons. For example: (a) repetition of content, e.g. three of the seven items on the physiological scale pertained to tiredness, and three of seven items on the fear of ageing subscale related to dementia; (b) difficult wording for a dichotomous response format; (c) poor relationship with the respective construct; (d) inclusion of worry topics; (e) inclusion of topics relating to work. A second group were inappropriately placed, e.g. physiological sensations (e.g. tension, restlessness, nervous energy, and keyed up) on the worry/oversensitivity subscale rather than on the physiological scale. A number of older adults included comments to the researcher about their problems constraining answers into a dichotomous format. Furthermore, dichotomous formats tend to cause an acquiescence bias (Saris, Revilla, Krosnick, & Shaeffer, 2010) and this could be exacerbated in older adults who may present with a degree of cognitive impairment.
Removal of the Lie Scale
One important finding from this study is that the inclusion of the Lie scale in the AMAS-E is problematic on the theoretical as well as statistical grounds. From a theoretical perspective the Lie scale includes items on social desirability (e.g. I am always kind) that represent age-appropriate social values, and do not elicit responses on suppression of anxiety. There is no evidence to suggest that this is causative of a low anxiety score, for example that people high on this scale would then conceal their anxiety. The inclusion of the Lie scale can not be justified on statistical grounds either, given the low second-order loading and the extremely low latent factor correlations with other subscales. Whilst the removal of the Lie scale did not increase the fit of the model, it is a more theoretically justified structure that is worthy of further examination.
The worry/oversensitivity subscale accounted for the most variance in scores, and had a very strong significant relationship with the total anxiety score (.93). It was also the longest subscale on the AMAS-E which may have overinflated the relationship with total anxiety compared to the other subscales. The physiological anxiety and fear of aging subscales had moderate significant relationships with the total anxiety scores (.70 and .73 respectively). The lie scale did not have a significant relationship with observed scores on the AMAS-E and was considered redundant.
The AMAS-E total score showed moderate convergence with the HADS anxiety (.56) and a weak relationship with the HADS depression scales (.30). AMAS-E worry/oversensitivity subscale showed a stronger relationship to the HADS anxiety subscale compared to the AMAS-E total score (.63 and .56 respectively) suggesting this may be a more effective measure of anxiety than the total score. The physiological anxiety and fear of aging subscales showed the weakest relationships with the HADS anxiety scale (.35 and .40 respectively), and interestingly the physiological subscale showed a nearly identical relationship to depression when compared to anxiety (.26 and .25 respectively). Potentially the high number of fatigue symptoms may have contributed to this. The lie scale had no relationship to anxiety or depression scores on the HADS.
Although the overlap between measures of anxiety and depression has been well documented (T. A. Brown, Campbell, Lehman, Grisham, & Mancill, 2001; Clark, Steer, & Beck, 1994; Wolitzky-Taylor et al., 2010), in this sample the HADS depression subscales showed a stronger relationship with the physiological anxiety subscale on the AMAS-E compared to the HADS anxiety subscale. This is indicative of poor discriminant validity of the AMAS-E, as somatic anxiety is considered to reflect pure anxiety within the tripartite model rather than the common negative affect factor (Clark et al., 1994).
The choice to exclude older adults aged over 80 means that the scale may not confidently be used in this population. Furthermore, the use of cognitive screening would have enabled further information on the validity of the scale in older adults, and potentially enable the explanation of variance in responses and evaluate the applicability of the AMAS-E to people with cognitive impairment. The sampled organisations consisted of predominantly New Zealand European women of middle to high socioeconomic status who were involved in community groups that valued active engagement in the community. Consequently the generalisability of findings to ethnic minority groups including Maori, and people with lower socioeconomic status was limited. Furthermore, the difference between the factor structure of responses on the AMAS-E between men and women was beyond the scope of the study due to the small sample size.
Summary of findings for the AMAS-E
The factor structure of the AMAS-E previously reported by Lowe and Reynolds (2000; 2006) was not strongly supported in an older adult sample, and a revision of the AMAS-E to remove the Lie scale while not improving model fit provides a more theoretically defensible measure. The removal of underperforming items may lead to an unacceptably small pool of items and consequent construct underrepresentation. In summary, the present study found limited validity evidence for the AMAS-E for older adults in its present format, and as such the use of the AMAS-E in older adult cohorts is cautioned.
Conflict of interest declaration:
Description of authors' roles
M. Roberts designed the study, collected and analysed the data, and wrote the paper. P. Merrick supervised the overall study and supported interpretation of the results and contributed to writing the paper. R. Fletcher contributed to the design of the study, choice and interpretation of statistical analysis.
Beck, A. T., Brown, G. K., Steer, R. A., & Weissman, A. N. (1991). Factor analysis of the dysfunctional attitudes scale in a clinical population. Psychological Assessment, 3, 478.
Brown, T. A. (2006). Confirmatory factor analysis for applied research. London: The Guilford Press.
Brown, T. A., Campbell, L. A., Lehman, C. L., Grisham, J. R., & Mancill, R. B. (2001). Current and lifetime comorbidity of the DSM-IV anxiety and mood disorders in a large clinical sample. Journal of Abnormal Psychology, 110(4), 585-599.
Byrne, B. (2001). Structural equation modeling with AMOS: Basic concepts, applications and programming. New Jersey: Lawrence Erlbaum Associates.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9(2), 233255. doi:10.1207/S15328007sem09025
Clark, D. A., Steer, R. A., & Beck, A. T. (1994). Common and specific dimensions of self-reported anxiety and depression: implications for the cognitive and tripartite models. Journal of Abnormal Psychology, 103(4), 645-654. doi:10.1037/0021-843X.104.3.542
Diefenbach, G. J., Stanley, M. A., & Beck, J. G. (2001). Worry content reported by older adults with and without generalized anxiety disorder. Aging & mental health, 5(3), 269-274. doi:10.1080/13607860120065069
Fan, X. T., Thompson, B., & Wang, L. (1999). Effects of Sample Size, Estimation Methods, and Model Specification on Structural Equation Modeling Fit Indexes. Structural Equation Modeling, 6(1), 56-83. doi:10.1080/10705519909540119
Heene, M., Hilbert, S., Draxler, C., Ziegler, M., & Buhner, M. (2011). Masking misfit in confirmatory factor analysis by increasing unique variances: A cautionary note on the usefulness of cutoff values of fit indices. Psychological Methods, 16(3), 319-336. doi:http://dx.doi.org/10.1037/ a0024917
Hu, L. T., & Bentler, P. M. (1998). Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychological Methods, 3 (4), 424-453. doi:10.1037//1082-989x.3.4.424
Lowe, P. A., & Reynolds, C. R. (2000). Exploratory analyses of the latent structure of anxiety among older adults. Educational and Psychological Measurement, 60(1), 100-116.
Lowe, P. A., & Reynolds, C. R. (2006). Examination of the psychometric properties of the Adult Manifest Anxiety Scale-Elderly Version scores. Educational and Psychological Measurement, 66(1), 93-115. doi:10.1177/0013164405278563
MacCallum, R. C., Widaman, K. F., Zhang, S. B., & Hong, S. H. (1999). Sample size in factor analysis. Psychological Methods, 4(1), 84-99. doi:10.1037//1082989x.4.1.84
Pachana, N. A., Byrne, G. J., Siddle, H., Koloski, N., Harley, E., & Arnold, E. (2007). Development and validation of the Geriatric Anxiety Inventory. International Psychogeriatrics, 19(1), 103-114. doi:10.1017/S1041610206003504
Preacher, K. J., & Coffman, D. L. (2006). Computing power and minimum sample size for RMSEA. Retrieved from http:// www.quantpsy.org/rmsea/rmsea.htm
Reynolds, C. R., Richmond, B. O., & Lowe, P. A. (2003). The Adult Manifest Anxiety Scale--Elderly Version (AMAS-E). Los Angeles: Western Psychological Services.
Roberts, M. H., Fletcher, R. B., & Merrick, P. L. (2014). The validity and clinical utility of the Hospital Anxiety and Depression Scale (HADS) with older adult New Zealanders. International Psychogeriatrics, 26(02), 325-333.
Saris, W. E., Revilla, M., Krosnick, J. A., & Shaeffer, E. M. (2010). Comparing questions with agree/disagree response options to questions with item-specific response options Symposium conducted at the meeting of the Survey Research Methods
Segal, D. L., June, A., Payne, M., Coolidge, F., & Yochim, B. (2010). Development and initial validation of a self-report assessment tool for anxiety among older adults: The Geriatric Anxiety Scale. Journal of Anxiety Disorders, 24, 709-714. doi:10.1016/j.janxdis.2010.05.002
Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the State-Trait Anxiety Inventory Palo Alto, CA: Consulting Psychologists Press.
Stanley, M. A., & Beck, J. G. (2000). Anxiety disorders. Clinical Psychology Review, 20(6), 731-754. doi:10.1016/S0272-7358(99)00064-1
Wetherell, J., Le Roux, H., & Gatz, M. (2003). DSM-IV criteria for generalized anxiety disorder in older adults: distinguishing the worried from the well. Psychology and Aging, 18(3), 622-627. doi:10.1037/088279126.96.36.1992
Wolitzky-Taylor, K. B., Castriotta, N., Lenze, E., Stanley, M. A., & Craske, M. G. (2010). Anxiety disorders in older adults: a comprehensive review. Depression and Anxiety, 27(2), 190-211. doi:10.1002/ da.20653
Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety and depression scale. Acta Psychiatrica Scandinavica, 67(6), 361-370.
Dr Margaret H Roberts, RN, DClinPsych.
School of Clinical Sciences Auckland University of Technology Private Bag 92006 North Shore Auckland New Zealand
+649 921 9999 extn 7711
Margaret H Roberts, Auckland University of Technology, Richard B Fletcher, Paul L Merrick, Massey University, Auckland
Caption: Figure 1. AMAS-E as a higher-order model with the complete item set (AMAS1)
Caption: Figure 2. Correlated AMAS-E model (AMAS2)
Table 1. Demographic information of participants N % Male 73 36 Female 117 57 Not Specified 13 7 Ethnicity New Zealand European 122 61 Other European 60 30 Maori 4 2 Not Specified 17 7 Highest Level Education Year 9 or 10 9 4.4 Year 11 or 12 25 12.3 University entrance 13 6.4 Tertiary 25 12.3 Post graduate 8 3.9 Not reported/ missing 120 59 Psychiatric History Any psychiatric history 22 10.8 Diagnosis of depression 13 6.4 Diagnosis of anxiety 5 5.7 Not reported/not elicited 112 55 Total participants 203 Table 2. Order of questionnaires for participants Package 1 Package 2 Package 3 Information Sheet Information Sheet Information Sheet Consent form Consent form Consent form HADS Trial measure AMAS-E AMAS-E HADS Trial Measure Trial measure AMAS-E HADS Table 3. Descriptive Statistics and Internal Consistency Coefficients for the AMAS-E AMAS-E Subscale N M SD a Interpretation of anxiety level Worry/ 167 6.36 5.40 .88 Normal Oversensitivity Physiological 184 2.00 1.99 .52 Normal Anxiety Fear of 183 2.61 2.13 .76 Normal Aging Lie 180 2.61 1.98 .71 Normal Total Anxiety 160 10.44 7.53 .88 Normal Table 4 Correlation Matrix Showing Convergent Validity between the AMAS-E and HADS HADS AMAS-E Anxiety Depression Total Worry/ Oversensitivity HADS Anxiety 1 Depression .56 * 1 AMAS Total .56 * .30 * 1 Worry/ .63 * .31 * .93 * 1 Oversensitivity Physiological .25 * .26 * .70 * .52 * Fear of aging .40 * .23 * .73 * .56 * Lie .05 -.01 .14 .11 AMAS-E Physiological Fear of Lie Aging HADS Anxiety Depression AMAS Total Worry/ Oversensitivity Physiological 1 Fear of aging .35 * 1 Lie .12 .21 * 1 N.B: * p <.01 Table 5. Goodness of Fit Indices for Competing Structural Models of the AMAS-E Model Df [CHI-SQUARE] TLI CFI RMSEA (90%CI) AMAS 1 898 1546.11 ** .72 .75 .059 (.054-.064) AMAS 2 896 1545.69 ** .72 .75 .059 (.054-.064) AMAS 3 626 1209.99 ** .70 .73 .067 (.062-.073) Model Mean Factor Loading AMAS 1 .51 AMAS 2 .51 AMAS 3 .49 AMAS1 NB: All factor loadings were significant except for A20, between Lie and Latent Anxiety, furthermore the second order factor loading was .08 for lie and latent anxiety. On the basis of this we correlated the factors and observed that the relationship between the anxiety factors and the lie scale was negative to low, that we removed the lie scale entirely. Table 6. Factor loadings between subscales of the AMAS1 and the hypothesised higher order factor of latent anxiety Latent Anxiety Worry/Oversensitivity 1.04 Physiological Anxiety .50 Fear of Aging .44. Lie .08 Table 7. Correlations between factors on AMAS2 model Worry/ Physiological Fear Lie Oversensitivity Anxiety of Aging Worry/ 1 Oversensitivity Physiological .52 (.53) 1 Anxiety Fear of Aging .46 (.46) .22 (.22) 1 Lie .08 -.00 .06 1 Values in brackets are the latent correlations with the Lie scale removed
|Printer friendly Cite/link Email Feedback|
|Author:||Roberts, Margaret H.; Fletcher, Richard B.; Merrick, Paul L.|
|Publication:||New Zealand Journal of Psychology|
|Date:||Nov 1, 2016|
|Previous Article:||Reducing racism against Maori in Aotearoa New Zealand.|
|Next Article:||Emotional labour in mental health field workers.|