| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychology, School of Philosophy, Psychology and Language Sciences, University of Edinburgh, Edinburgh, Scotland (B.A.S., M.D.T., I.J.D.); and the MRC Social and Public Health Sciences Unit, University of Glasgow, Glasgow, Scotland (G.D.).
Address correspondence and reprint requests to Ian J. Deary, PhD, FRCPE, Psychology, School of Philosophy, Psychology and Language Sciences, University of Edinburgh, 7, George Square, Edinburgh EH8 9JZ, Scotland. E-mail ian.deary{at}ed.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSCONCLUSIONSNOTESREFERENCES |
|---|
Methods: Participants were from the UK Health and Lifestyle Survey (HALS), a national sample survey of adults in England, Scotland, and Wales. In 1984/1985, data on lifestyle factors, socioeconomic status, and health were collected for 9,003 people. Cognitive data were collected for 7,414 individuals. All-cause mortality was investigated over 19 years of follow-up in relation to simple and choice reaction time, performance on a short-term verbal declarative memory test, and on a test of visual-spatial reasoning.
Results: Slower and more variable simple and choice reaction times were significantly related to increased risk of all-cause mortality over 19 years of follow-up. The increased risk of all-cause mortality was partly attenuated after adjustments for socioeconomic status, health behaviors, and health status. A novel finding was the existence of an effect of reaction time on all-cause mortality in young adults. Poorer verbal memory ability was also significantly related to an increased risk of dying in young adults independently of reaction time score.
Conclusion: Slower and more variable reaction time was related to higher mortality risk in younger as well as older participants. Among younger adults, higher memory ability was also associated with lower risk of dying. The cognition-mortality relationship may be explained in part by the brain’s efficiency of information processing and memory performance.
Key Words: cognition • mortality • information-processing speed • social position
Abbreviations: HALS = Health and Lifestyle Survey; FEV = forced expiratory volume; BMI = body mass index.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSNOTESREFERENCES |
|---|
To date, we know little of which specific aspects of cognitive functioning relate to health and mortality because most studies have used group-based measures of general intelligence (e.g., 7,13). Such omnibus tests do not afford an understanding of the particular cognitive domains or processes that might link mental ability differences with inequalities in survival. More recent studies have begun to use measures such as reaction time indices that could be considered more fundamental indicators of the brain’s processing efficiency. Their items are simple and identical, and they afford the derivation of timed estimates of subjects’ processing efficiency and consistency.
With this in mind, Deary and Der (13) examined a sample of 898 middle-aged adults (aged 54–58 years) from the West of Scotland Twenty-07 Study. Participants were followed up for 14 years. They found strong significant associations between mortality and baseline performance on the Alice Heim 4 Part 1 test of general intelligence (hazard ratio = 1.38), mean simple reaction time (hazard ration = 1.30), and mean choice reaction time (hazard ratio = 1.37), even after adjustment for sex, education, occupational social position, and smoking. Importantly, in multivariate analyses in which the Alice Heim and reaction time tests were both entered as predictors, the Alice Heim test failed to make a significant, independent contribution to mortality, indicating that the relatively simple index of reaction time might account for the effect of IQ on mortality.
These findings on psychometric intelligence and reaction time and mortality should be viewed in light of the mechanisms that have been proposed to explain the association between cognitive ability and all-cause mortality (12). First, higher cognitive ability is associated with more favorable social circumstances; for example, through being a member of a higher social class or achieving better educational qualifications (14), which are, in turn associated with lower mortality (12). Second, higher cognitive ability may be linked to behaviors that are conducive to good health such as healthy eating, a low alcohol intake, and avoidance of smoking (12). Third, low cognitive ability may be a proxy indicator of deficient brain development that is correlated with later adulthood illness. For example, low childhood cognition has been found to be an indicator of a poor prenatal environment or early life insults such as malnutrition and low birthweight, which are associated with an increased risk of cardiovascular disease in later life (15–17). Finally, and especially relevant here, psychometric intelligence may assess general bodily integrity by indexing efficiency of information processing within the central nervous system (7). In the analysis of the Twenty-07 study (13), the effect of psychometric general intelligence on mortality was nonsignificant once reaction time was controlled for. These results suggest that a reduced efficiency of information processing may account for much of the association between low IQ and earlier death. Therefore, as a more culture-reduced measure of information processing, reaction time may provide a sensitive indicator of the body’s clinical pathology.
Previous studies have shown an association between reaction time and health. Pavlik et al. (18) found a significant association between the combination of hypertension and type 2 diabetes and worse simple reaction time performance. Anstey et al. (19) found that pulmonary function, as measured by forced expiratory volume (FEV) at 1 second, is positively associated with measures of speed. Deary and Der’s (13) analysis of the Twenty-07 study is the only study to date that has examined the influence of reaction time mean and variability on mortality in a representative sample of the population. Its suggestion that reaction time may be a relatively sensitive indicator of risk of mortality risk needs replicating and extending. The report was based on middle-aged adults only. It is important to determine whether the association between reaction time and mortality is also present in at other ages in adulthood. It is especially important to examine younger age groups, in whom any association between reaction time and mortality is less likely to be explained by pre-existing (but perhaps subclinical) illness in some participants. If the association between reaction time and mortality exists in young adults, this would indicate that reaction time denotes more than mere physiological deterioration with age.
Therefore, the principal aim of the present study is to examine the association between reaction time indices and mortality in a large representative sample that covers the whole adult age range. Additionally, with data on two further types of cognitive measures (verbal declarative memory and visual-spatial reasoning, both measured in brief tests), we can examine the effects of individual differences in these specific cognitive domains on mortality.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSNOTESREFERENCES |
|---|
Initial sample selection and interviewing took place between 1984 and 1985, where a total of 12,254 addresses were randomly chosen from UK Electoral Registers. One individual aged 18 years or over was chosen from each household, yielding interviews with 9,003 individuals aged between 18 and 99 (21). Physical measurements, including reaction times, were available for 7,414 of these. All participants are continuously followed up for mortality and cancer registration. The latest follow-up was completed in May 2003. Full details of the design and sampling have been described elsewhere (20).
The study population was compared with the 1981 census to examine whether it could be assumed to represent the general population (21). The survey contained more women than the general population, which is likely to be due to availability for interview. Single people and those who were divorced/separated were underrepresented. People in the lower socioeconomic groups in terms of income and education were also less likely to complete baseline measurement. These biases are small, and the survey provides a reasonably good representation of the general adult population (21).
Procedure
Information was collected from two home visits (21). The first was carried out by an interviewer and lasted approximately 1 hour. The data collected included information on home and family circumstances, educational attainment, income, self-reported health, health attitudes, diet, leisure, work, exercise, smoking status, and alcohol consumption.
In the second home visit, a nurse completed a physiological examination of the participant. This involved taking measurements of height, weight, girth, blood pressure, pulse rate, respiratory function, and environmental and exhaled carbon monoxide. A brief cognitive battery comprising tests of short-term verbal declarative memory, visual-spatial reasoning, and reaction time was administered during this visit (20–22).
Measures
Reaction Time
Participants were assessed on simple and four-choice reaction time using a portable battery-operated device specially designed for the study (23). A schematic diagram of the device is shown in Figure 1. Simple reaction time was the time taken for the subject to press a key as quickly as possible after being presented with a stimulus. The participant rests a finger of their preferred hand on the button marked 0 and presses it as soon as a 0 appears on the screen. Eight practice trials were presented before 20 test trials. Mean reaction time and SD were recorded in milliseconds for each participant.
|
Choice reaction time was the time taken to press one of four keys corresponding to one of four digits (1–4) presented on the screen. The participant rests two fingers of each hand on the buttons marked 1, 2, 3, and 4 and presses the button that matches the digit presented in the display. Eight practice trials are followed by 40 test trials in which each digit is randomly presented 10 times at an interval of between 1 and 3 seconds. Mean reaction time and SD in milliseconds are calculated separately for correct and incorrect responses, and the number of errors was also recorded for each participant.
Results presented here for choice reaction time are based on correct trials only. All reaction time measures were standardized to zero mean and unit SD.
Verbal Declarative Memory
A list of 10 common foods (roast meat, digestive biscuits, potatoes, eggs, orange juice, grilled fish, Weetabix, white bread, cheese, apples) was read out to the respondent in the context of a discussion about the fiber content of food. Participants were initially asked to report which foods contained dietary fiber. A few minutes later, and without prior warning, participants were asked to recall the same 10 food items. This provided a measure of "incidental" verbal memory.
Visual-Spatial Reasoning
A block counting test was used as a measure of visual-spatial reasoning. Participants were presented with a set of six two-dimensional pictures that represented three-dimensional piles of blocks and were required to calculate the number of blocks used to make each pile. Calculation of the correct answer was made by making inferences about the piles and counting both the blocks that could be seen and those that were hidden from view. The total number of correct responses was recorded, with a range of 0 to 6. An example of this type of item is provided in Figure 2.
|
Demographic and Lifestyle Factors
Age
Age was calculated as age in years at interview in 1984/1985.
Social Class
Social class was derived using the UK Registrar General’s six-fold classification (24): professional (I), managerial (II), skilled nonmanual (IIIN), skilled manual (IIIM), semiskilled manual (IV), and unskilled (V). Household social class was used in this analysis and was based on the current occupation of the "head of household" or the past or "usual" occupation of the retired or unemployed respondents.
Education
Education was recorded as the highest qualification obtained either while at school or gained since school. This variable had five levels: first or higher degree; semiprofessional or professional qualification, for example nursing or teaching qualification; A-level or equivalent, including advanced City and Guilds certificates; O-level or equivalent, including ordinary City and Guilds certificates; no educational qualifications, work-related certificates were included in this category.
Smoking Status
Current smoking status was recorded as never smoked, ex-smoker, current occasional smoker, and current regular smoker. An occasional smoker was defined as smoking less than one cigarette per day, whereas a regular smoker was defined as smoking one or more than one cigarette per day.
Alcohol Consumption
Two measures of alcohol consumption were used. Level of alcohol consumption per week was measured in alcohol units. One unit of alcohol was equivalent to half a pint of beer or cider, or a single measure of spirits, or a glass of wine, or a small glass of fortified wine. Current alcohol consumption was recorded as nondrinker, very special occasion drinker, occasional drinker, and regular drinker.
Physical Activity
Information on a variety of physical activities (work-related activity, walking, housework, gardening, DIY, and sports activities) was obtained during the interview. For some activities, the number of occasions was recorded, and for others the duration was recorded in minutes. In both cases, the intensity of the activity was also recorded. We have aggregated the information to produce four physical activity variables: minutes spent doing vigorous activity, occasions spent doing vigorous activity, minutes spent doing nonvigorous activity, and occasions spent doing nonvigorous activity. Vigorous activity was defined as activity that caused breathlessness.
FEV
FEV in 1 second was used as a measure of respiratory function. It was measured using a Micro Medics electronic spirometer and divided by the height in meters squared to take account of differences in body size.
Blood Pressure
Blood pressure was measured with an Accutorr sphygmomanometer. Four measures of resting systolic and diastolic blood pressure were taken at 1-minute intervals. The median value of both systolic and diastolic blood pressure was used for this analysis.
Body Mass Index (BMI)
BMI (kg/m2) was derived from height, measured in meters with a portable stadiometer, and weight in kilograms measured using electronic scales.
Vital Status
Deaths of survey members are continually notified by the National Health Service central registry, which also provided copies of death certificates. The data are periodically updated with the information provided, and the data used here comprise the latest available update of May 2003. Death certificates contained information on up to three primary and two secondary causes of death coded according to the Ninth International Classification of Diseases (25). For the present analysis, a single "underlying cause" of death was used.
Statistical Analysis
For the mortality analyses, survival time was age at death or age at May 2003 for surviving participants. Cox’s proportional hazards regression, as implemented in the SAS Version 9.1 Phreg procedure (26), was used to calculate hazards ratios. These provided estimates of the proportionate change in the mortality risk for each SD change in reaction time parameters and for a 1-point change in verbal declarative memory or visual-spatial reasoning. Adjustments were made for covariates, including social class, education, smoking status, alcohol consumption, physical activity, FEV, blood pressure, and BMI as these have been related both to mortality risk (27) and cognitive ability (14,19,28–32). The control variables were entered in blocks, in a hierarchical set of analyses, as follows:
Hazard ratios, 95% confidence intervals, and p values were used for interpretation. The direction of the verbal declarative memory and visual-spatial reasoning scores was opposite to the reaction time scores as a high score on the former two tests and a low score in the latter’s parameters represented better cognitive ability. For example, for reaction time a value of 1.02 would be interpreted as a 2% increase in mortality risk for every SD increase in reaction time, whereas for verbal declarative memory, a hazard ratio of 0.98 would mean that for every point increase in memory score, risk of mortality decreases by 2%.
Preliminary analyses showed that the hazard was not proportional with age, as is assumed by the model. Therefore, we present results both for all ages together and for three broad age groups. The continuous age variable included 18- and 19-year-olds (n = 222). However, they were dropped when three age groups were constructed (20–39 years, 40–59 years, 60+ years) to limit the groups to 20-year age bands. The age groups allowed us to examine whether the relationship between cognition and mortality was comparable across different age bands.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSNOTESREFERENCES |
|---|
As expected (33) Spearman’s rank correlations showed small but highly significant negative correlations between the memory and visual-spatial reasoning tests and the reaction time measures (Table 1). The largest correlations between these measures were seen in the 60+ age group.
|
Simple Reaction Time
When all ages were examined together, there was a highly significant effect of simple reaction time mean on risk of dying in the age- and sex-adjusted baseline model (HR = 1.09; Table 2). The size of this effect was somewhat attenuated when social class and education were added to the SES model (HR = 1.06). Further attenuation was seen in the health behaviors models (HR = 1.04) and 4 (HR = 1.03). The effect of simple reaction time variability was of the same magnitude in the baseline age and sex model (HR = 1.09; Table 2). The effect was slightly attenuated when social class and education were entered into the SES model (HR = 1.08). The largest attenuation was seen in the health status model (HR = 1.03).
|
When the sample was split into three age bands, there was a significant effect of simple reaction time mean on risk of mortality after controlling for age and sex in the 20- to 39-year age band (HR = 1.39) and in those over 60 years (HR = 1.09; data tables available from authors). The effect decreased slightly in the SES model to 1.34 and 1.07, respectively. There was little further attenuation with the addition of the health behavior block of variables (smoking status, alcohol consumption, and physical activity). However, the effect decreased further in the 20- to 39-year age group in the health status model (HR = 1.26). A significant effect of simple reaction time variability on mortality was noted in the 60+ age group only (HR = 1.09; data available from authors). Further adjustment for social class and educational attainment altered the effect size slightly (HR = 1.07). There was further slight attenuation in the health behaviors models (HR = 1.05) and 4 (HR = 1.03). All results remained highly significant. The hazard ratios for simple reaction time variability were similar in the baseline, SES, and health behaviors models across the three age groups but did not reach statistical significance in the younger two groups owing to the smaller numbers of deaths.
Choice Reaction Time
After adjusting for age and sex, there was a highly significant effect of choice reaction time mean on mortality in the overall sample (HR = 1.18; Table 2). This effect was slightly attenuated on adjustment for social class and education (HR = 1.15). Further attenuation was noted in the health behaviors model (HR = 1.10) and 4 (HR = 1.08). The effect of choice reaction time variability on mortality across the entire sample was also highly significant in the baseline age and sex model (HR = 1.08; Table 2). Again, the effect attenuated in model 2 (HR = 1.06). Adjustment for the health behavior block (smoking status, alcohol consumption, and physical activity) further reduced the size of the effect to 1.04.
With the sample split into three age bands, the association between choice reaction time mean and mortality controlling for sex and age was significant for all three age groups (20–39: HR = 1.62; 40–59: HR = 1.20; 60+: HR = 1.17; data tables available from authors). After adding social class and education, only the 20- to 39-year age group (HR = 1.59) and 60+ age group (HR = 1.15) showed significant effects. Further attenuation was noted for these two age groups in the health behaviors models (20–39: HR = 1.56; 60+: HR = 1.11) and 4 (20–39: HR = 1.41; 60+: HR = 1.09). For choice reaction time variability, a significant association with mortality was noted in the baseline sex and age model in the 20- to 39-year age group (HR = 1.66) and the 60+ age group (HR = 1.07). The effect sizes were slightly attenuated in the 20- to 39-year age group on the addition of social class and education to the model (HR = 1.60). In the 60+ age group, the effect sizes were altered only a little with the addition of these variables to the model (HR = 1.06). Health behaviors (HR = 1.57) and health status (HR = 1.50) models showed significant effects in the 20- to 39-year age group only. The effect size for the 40- to 59-year age group was larger than that of the 60+ age group but did not reach statistical significance owing to the smaller number of deaths.
Verbal Declarative Memory
When all ages were examined together, there was a highly significant effect of memory performance on risk of dying after controlling for age and sex (HR = 0.91, 95% CI, 0.88–0.94; Table 2). Adding social class and education to the model slightly decreased the strength of the association. Further attenuation was noted when the health behavior variables (HR = 0.94) and the physical health variables (HR = 0.95) were added to the model.
After controlling for age and sex, lower memory performance was significantly associated with mortality risk in the 40- to 59-year age band (HR = 0.90) and in those aged 60+ (HR = 0.90). On adding social class and education, the only significant association between lower memory and mortality risk was seen in those aged 60+ years (HR = 0.92). Further attenuation was noted in health behaviors (HR = 0.94) and health status (HR = 0.94) models. There was no significant association between memory performance and mortality risk in the 20- to 39-year age group even before adding all of the control variables to the model. However, the hazard ratio in this group was the same as that found in the 60+ age group; it failed to reach statistical significance, which is likely to be due to the small number of deaths in the youngest group (data tables available from authors).
Visual-Spatial Reasoning
When all ages were examined together, the only significant association between visual-spatial reasoning and risk of dying was in the baseline model (HR = 0.95; Table 2).
With the sample split into three age bands, the only significant association was noted in the baseline model in the 60+ age group (HR = 0.96). There was no significant association between visual-spatial reasoning and risk of dying in the 20- to 39-year age group (even though the effect size in this group was the largest, it was nonsignificant owing to the small number of deaths) and 40- to 59-year age group even before adding all of the variables to the model (data tables available from authors).
Backwards Elimination Variable Selection
Backwards elimination was used to determine the best predictors of mortality for the entire sample and separately for each age group. This was completed from a model containing all six cognitive measures and controlling for the demographic, socioeconomic, health behavior, and physical health variables. For all ages together, the model selected included memory and choice reaction time mean. For the 20- to 39-year age group, the model selected included only choice reaction time mean. For the 40- to 59-year age group, none of the six cognitive measures were selected. Finally, the model for the 60+ age group included only choice reaction time mean.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSNOTESREFERENCES |
|---|
Whereas the effects that were statistically significant were identified for further comment and analyses, this should not obscure the fact that some of the effect found in the younger groups had a similar effect size to those in the older group, even though they were often not significant at p < .05. This is a consequence of there being fewer deaths in the younger groups. The frequent parity of the effect size across age groups should encourage larger studies with younger samples.
The considerably larger effect of reaction time mean and variability on mortality in the 20- to 39-year age group compared with the 60+ age group is a surprising finding. We considered the possibility that there might have been different causes of death per age group; for example, accidents in the 20- to 39-year age group. However, this was not the case as the most common specific causes of death for all three age groups were cancer and cardiovascular disease (Table 3). Both simple and choice reaction time mean and variability were stronger predictors of mortality than memory. However, after further adjustment for all reaction time scores, the influence of memory on mortality remained in the 60+ age group (full results not shown; HR = 0.94; 95% CI, 0.90–0.97; p = .0007). Visual-spatial reasoning showed very little effect on mortality.
|
The effect of reaction time mean and variability on mortality in the 40- to 59-year age group and the 60+ age group was different from the one previous study that examined this association (13). We found a small effect of reaction time mean and variability on mortality in the 60+ age group, but, though the effect sizes were similar, we tended not to find significant associations in the 40- to 59-year age group. Deary and Der’s (13) study found highly significant effects among those aged between 54 and 58 years even after controlling for education, social class, deprivation, and smoking. The size of the effect noted in the 60+ age group was smaller than previously noted. We obtained hazard ratios that ranged between 1.05 and 1.12 for each of the four reaction time measures, whereas Deary and Der (13) found effect sizes of between 1.15 and 1.37 for the same measures and controlling for the abovementioned variables. The smaller effects noted in the present study may be explained by the occurrence of natural death as this age group contained some much older participants, compared with Deary and Der’s (13) very narrow age range. However, it might also be possible that the present study is a more accurate indicator of the relationship between reaction time and mortality across mid to late adulthood.
The independent effect of memory performance on mortality seen here in the 60+ age group has been noted in previous longitudinal studies (36,37). However, other authors have found no relationship (6,38,39). Reasons for possible discrepancies in findings may be attributed to differences in length of follow-up, the use of different tests of memory ability, the addition of different mediators into the model, and the brevity of the cognitive tests used in the present study. The lack of an effect in the 40- to 59-year age group is inconsistent with work completed by Pavlik et al. (11) using data from the large Atherosclerosis Risk in Communities Study. However, this study has a considerably shorter follow-up period (6.3 years) than the present study’s 19 years.
How does cognitive ability predict mortality? One way of explaining this is through the trait-state dichotomy. Seeing cognitive ability as a state, the "common cause" hypothesis posits that the brain is merely one part of a deteriorating body where the existence of a common factor reflects age-related deterioration in cognitive and biological processes (40). Therefore, reaction time may be one way of disclosing this general bodily deterioration (41). However, the fact that we found a strong association between reaction time mean and variability and mortality in a young sample that would be relatively free of age-related degeneration when first tested suggests a more trait-like portrayal of cognitive ability. In this case, perhaps reaction time is picking up some aspects in the brains of even initially healthy individuals that may be associated with survival.
Although the effect was weaker, memory ability predicted mortality after reaction time was controlled for. This finding is of interest because it supports the theory that certain psychometric factors show stronger relationships with mortality than others (42,43).
The results suggest that reaction time mean and variability and cognitive function are related to all-cause mortality. The partial attenuation of the cognition-mortality association by social, demographic, health behaviors, and physical health factors suggests that both memory ability and reaction time explain part of this association in not only older age samples but also among the young. These results suggest that reaction time is not merely an indicator of age-related physiological deteriorations but rather an indicator of the brain’s more basic information processing ability, suggesting that slower and more variable processing skills are a risk factor for mortality in themselves.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSNOTESREFERENCES |
|---|
Received for publication November 24, 2004; revision received September 1, 2005.
DOI:10.1097/01.psy.0000195867.66643.0f
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSCONCLUSIONSNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  N K J Oksala, H Jokinen, S Melkas, A Oksala, T Pohjasvaara, M Hietanen, R Vataja, M Kaste, P J Karhunen, and T Erkinjuntti Cognitive impairment predicts poststroke death in long-term follow-up J. Neurol. Neurosurg. Psychiatry, November 1, 2009; 80(11): 1230 - 1235. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Jokela, G. D. Batty, I. J. Deary, C. R. Gale, and M. Kivimaki Low Childhood IQ and Early Adult Mortality: The Role of Explanatory Factors in the 1958 British Birth Cohort Pediatrics, September 1, 2009; 124(3): e380 - e388. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. A. Hall, J. A. Dubin, M. Crossley, M. E. Holmqvist, and C. D'Arcy Does Executive Function Explain the IQ-Mortality Association? Evidence from the Canadian Study on Health and Aging Psychosom Med, February 1, 2009; 71(2): 196 - 204. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. W. Roberts, N. R. Kuncel, R. Shiner, A. Caspi, and L. R. Goldberg The Power of Personality: The Comparative Validity of Personality Traits, Socioeconomic Status, and Cognitive Ability for Predicting Important Life Outcomes Perspectives on Psychological Science, December 1, 2007; 2(4): 313 - 345. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. A. Shipley, G. Der, M. D. Taylor, and I. J. Deary Association Between Mortality and Cognitive Change Over 7 Years in a Large Representative Sample of UK Residents Psychosom Med, September 1, 2007; 69(7): 640 - 650. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. W. Rebok, M. C. Carlson, and J. B. S. Langbaum Training and Maintaining Memory Abilities in Healthy Older Adults: Traditional and Novel Approaches J Gerontol B Psychol Sci Soc Sci, June 1, 2007; 62(Special_Issue_1): 53 - 61. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
