Semipartial Correlation Research Articles

Cardiovascular Fitness Is Negatively Associated With Homocysteine Levels in Female Adolescents

To examine the association between cardiovascular fitness and homocysteine levels in adolescents. Cross-sectional study. Madrid, Murcia, Granada, Santander, and Zaragoza, Spain. One hundred fifty-six Spanish adolescents (76 boys and 80 girls) aged (mean +/- SD) 14.8 +/- 1.4 years. Cardiovascular fitness was measured by the 20-m shuttle run test. Pubertal stage, birth weight, smoking status, and socioeconomic status were determined, and the sum of 6 skinfold thickness measurements, and serum folic acid and vitamin B(12) levels were measured. Methylenetetrahydrofolate reductase (MTHFR; 677C>T genotype) polymorphism was done by DNA sequencing. Fasting homocysteine levels. Mean values of homocysteine were significantly higher in the MTHFR 677CT and TT genotype subgroups compared with the CC genotype subgroup in adolescent boys, whereas in adolescent girls, mean values of homocysteine were significantly higher in the MTHFR 677CT and TT genotype subgroup compared with the CC and CT genotype subgroups. Multiple regression analyses showed that cardiovascular fitness was significantly associated with homocysteine levels in female adolescents after controlling for potential confounders including the MTHFR 677C>T genotype (beta = -0.40; semipartial correlation = -0.35; P = .007). No associations were found between cardiovascular fitness and homocysteine levels in male adolescents (beta = 0.12; semipartial correlation = 0.08; P = .51). The results suggest that cardiovascular fitness is negatively associated with homocysteine levels in female adolescents after controlling for potential cofounders including MTHFR 677C>T genotype.

Is Exposure to Media Intended for Preschool Children Associated With Less Parent-Child Shared Reading Aloud and Teaching Activities?

To determine whether electronic media exposure is associated with decreased parental reading and teaching activities in the homes of preschool children. A convenience sample presenting for well-child care to an urban hospital pediatric clinic was enrolled. Inclusion criteria were: child's age 3 to 5 years and not yet in kindergarten. Electronic media exposure (TV, movies/video, computer/video games) was assessed with a 24-hour recall diary and characterized on the basis of industry ratings. Reading aloud and teaching activities were assessed with the StimQ-Preschool READ and PIDA (Parental Involvement in Developmental Advance) subscales, respectively. A total of 77 families were assessed. Children were exposed to a mean (SD) of 200.8 (128.9) minutes per day of media, including 78.2 (63.7) minutes of educational young child-oriented, 62.0 (65.6) minutes of noneducational young child-oriented, 14.8 (41.4) minutes of school age/teen-oriented, and 29.2 (56.6) minutes of adult-oriented media, as well as to 16.6 (47.5) minutes of media of unknown type. A total of 79.2% watched 2 or more hours per day. Noneducational young child-oriented exposure was associated with fewer reading (semipartial correlation coefficient [SR] = -0.24, P = .02) and teaching (SR = -0.27, P = .01) activities; similar relationships were not found for other media categories. Children exposed to 2 or more hours of total electronic media per day had 1.6 (95% confidence interval, 0.4-2.9) fewer days per week of reading than children exposed to less than 2 hours (SR = -0.27, P = .01). This study found an association between increased exposure to noneducational young child-oriented media and decreased teaching and reading activities in the home. This association represents a mechanism by which media exposure could adversely affect development.

Testing Complex Correlational Hypotheses With Structural Equation Models

It is often of interest to estimate partial or semipartial correlation coefficients as indexes of the linear association between 2 variables after partialing one or both for the influence of covariates. Squaring these coefficients expresses the proportion of variance in 1 variable explained by the other variable after controlling for covariates. Methods exist for testing hypotheses about the equality of these coefficients across 2 or more groups, but they are difficult to conduct by hand, prone to error, and limited to simple cases. A unified framework is provided for estimating bivariate, partial, and semipartial correlation coefficients using structural equation modeling (SEM). Within the SEM framework, it is straightforward to test hypotheses of the equality of various correlation coefficients with any number of covariates across multiple groups. LISREL syntax is provided, along with 4 examples.

The “hidden” semantic category dissociation in mild-moderate Alzheimer's disease patients

In patients manifesting mild-moderate Alzheimer's disease (AD), lexical semantic tasks are known to be influenced by several variables which should be adequately taken into account when studying semantic category dissociations. The following study provides indexes of three new variables (imageability ( I), percentage of name agreement (pNA) and number of target alternatives (nTA)) and investigates their role in naming in a group of people with AD and in matched older adults controls. Forty young healthy participants rated I, pNA and nTA of 155 stimuli (including living and non-living items) from Laiacona et al. (1993) and Sartori et al. (1993) sets. Forty-eight people with mild-moderate AD and 40 older adults were given the two naming tests and their naming ratings were analysed with a two-way ANOVA (two groups × two categories) to assess category specificity and the effect of interaction. The influence of relevant concomitant variables in naming was measured using a multiple regression analysis. Semi-partial correlations were carried out to assess the independent contribution of each variable to naming. We found that living items were more imageable and had fewer lexical alternatives and higher name agreement than non-living items. We also found that controls significantly named better than AD patients ( F = 37.551, p < .001), whilst the two-way ANOVA showed no significant effect of category ( F = .649, p = .423). Notably category effect emerged when assessing its independent contribution performing a semi-partial correlation ( β = −.278, p < .001) which kept the effect of relevant concomitant variables under control. Our results confirm that category dissociation does emerge in mild-moderate AD patients when the effect of relevant concomitant variables is adequately taken into account. The hypothesis that the highly correlated properties of items from biological categories may play a protective effect on living things, making them less prone to impairment in the early stages of AD, is discussed.

Determinants of repeated-sprint ability in females matched for single-sprint performance

This study investigated the relationship between VO2max and repeated-sprint ability (RSA), while controlling for the effects of initial sprint performance on sprint decrement. This was achieved via two methods: (1) matching females of low and moderate aerobic fitness (VO2max: 36.4 +/- 4.7 vs 49.6 +/- 5.5 ml kg(-1) min(-1) ; p < 0.05) for initial sprint performance and then comparing RSA, and (2) semi-partial correlations to adjust for the influence of initial sprint performance on RSA. Tests consisted of a RSA cycle test (5 x 6-s max sprints every 30 s) and a VO2max test. Muscle biopsies were taken before and after the RSA test. There was no significant difference between groups for work (W1, 3.44 +/- 0.57 vs 3.58 +/- 0.49 kJ; p = 0.59) or power (P1, 788.1 +/- 99.2 vs 835.2 +/- 127.2 W; p = 0.66) on the first sprint, or for total work (W(tot), 15.2 +/- 2.2 vs 16.6 +/- 2.2 kJ; p = 0.25). However, the moderate VO2max group recorded a smaller work decrement across the five sprints (W(dec), 11.1 +/- 2.5 vs 7.6 +/- 3.4%; p = 0.045). There were no significant differences between the two groups for muscle buffer capacity, muscle lactate or pH at any time point. When a semi-partial correlation was performed, to control for the contribution of W1 to W(dec), the correlation between VO2max and W(dec) increased from r = -0.41 (p > 0.05) to r = -0.50 (p < 0.05). These results indicate that VO2max does contribute to performance during repeated-sprint efforts. However, the small variance in W(dec) explained by VO2max suggests that other factors also play a role.

Books, Toys, Parent-Child Interaction, and Development in Young Latino Children

To describe the interrelationships between books and toys in the home, parent-child interaction, and child development at 21 months among low-income Latino children. Latino mother-infant dyads enrolled in a level 1 nursery and infants were followed to 21 months. The subjects consisted of the control group of a larger intervention study. At 6 and 18 months, the number of books and toys in the home and the frequency of reading aloud were measured by the StimQ. At 21 months, child cognitive and language development and parent-child interaction were assessed by the Bayley Mental Development Index (MDI), the Preschool Language Scale-3 (PLS-3), and the Caregiver-Child Interaction Rating Scale, respectively. Eligibility for early intervention (EI) services was determined on the basis of the MDI and PLS-3. Data were obtained for 46 (63.0%) of 73 at 21 months. In multiple regression analysis, books provided at 18 months predicted both cognition (semipartial correlation [sr] = .49, P= .001) and receptive language (sr = .37, P= .02), whereas toys provided at both 6 and 18 months predicted 21-month receptive language (sr = .40, P= .01; sr = .32, P= .047, respectively). Reading aloud by parents > or =4 days a week was associated with decreased EI eligibility (adjusted odds ratio = 0.16, 95% confidence interval 0.03-0.99). Reading aloud and provision of toys are associated with better child cognitive and language development as well as with decreased likelihood of EI eligibility.

Seagrass complexity hierarchies: Influence on fish groups around the coast of Jersey (English Channel)

Recent research has identified a need for seagrass habitat management plans to be based on landscape-level approaches as they offer a more appropriate scale for large mobile fauna than smaller scales. Also, conservation decisions are more likely to be a choice between different seagrass beds rather than parts of individual beds. The present study examined the spatial utilisation of subtidal seagrass beds by fish around the coast of Jersey, English Channel (49°N 02°W) with the aim of identifying influential scales of complexity. A hierarchical-scale of landscape configuration and habitat characteristics was measured for eight seagrass beds using aerial photographic analysis (e.g. core area, contiguity and other landscape metrics), digital echo-sounder data (e.g. depth, canopy height) and diver surveys (e.g. epiphyte load). The contributions of these variables as predictors of functional fish groups were explored using multiple linear regression models. Results indicated that more fragmented seagrass beds supported lower numbers of fish species than more homogenous seagrass landscapes (squared semi-partial correlation coefficient, sr 2 = − 0.3). Densities of juveniles of larger fish species showed a negative relationship with increased fragmentation of the seagrass (sr 2 = − 0.34). At smaller scales of structural complexity the densities of cryptic fish were related positively to canopy height (sr 2 = 0.46). At night, fewer patterns could be explained by the independent variables in the model, which was attributed to the greater movement of fish between the seagrass and adjacent habitats to forage, and a breakdown in the association with seagrass habitat as a refuge from predation.

Publication Rates of Male and Female Academic Clinical Psychologists in California

Compared here was the publication rate of men and women surveyed by mail (480 academic clinical psychologists in California) regarding number of publications, self-perception of their role of researcher versus teacher, variables related to research opportunity (grants, time spent in research), and type of institution in which they are employed. The men had 1.6 as many publications per year as did the women, and the semipartial correlation was -.21 for sex in the last step of a multiple regression and accounted for an additional 4% of the variance. Thus, the men published more, even when ostensible opportunities were controlled. For this sample variables of sex and self-perception of role of researcher versus teacher were more highly correlated with number of publications than were the research opportunity variables of grants, time spent in research, and type of academic institution of employment.

Self-Report and Pain Behavior Among Patients With Chronic Pain

To determine the relationship between pain behaviors and self-report of pain and disability in patients with chronic pain. Thirty-nine patients (59% women), aged 19 to 79 years, admitted to a Multidisciplinary Pain Center with chronic noncancer pain, were assessed on the following: 1) pain intensity using a 0 to 10 Numerical Graphic Rating Scale; 2) the Pain Disability Index; 3) the Pain Self-Efficacy Questionnaire; 4) a 10-minute videotaped session involving sitting, standing, walking, and reclining, after which pain behaviors were coded using a standardized observational protocol. Interrelationships among pain intensity, pain disability, self-efficacy, and pain behavior were tested using the Pearson product-moment correlations. Significant relationships were found between pain behavior frequencies and self-reported pain intensity (r = 0.29, P < 0.05), self-reported pain disability (r = 0.54, P < 0.0005), and reported self-efficacy (r = -0.42, P < 0.005). Multiple regression analyses were performed to further investigate these interrelationships. Only the Pain Disability Index score was found to make a significant unique contribution (semipartial correlation of 15%, P < 0.008) to the prediction of total pain behavior score. Findings suggest that pain behavior observation is a valid and reliable assessment tool for use with a heterogeneous chronic pain population. Significant associations were found between pain behaviors and self-report measures of pain intensity, pain disability, and self-efficacy; pain intensity scores displayed a weak relationship; and pain disability scores the strongest relationship with pain behavior. Viewed with previous research, the results of this study indicate the value of a multimodal, cognitive-behavioral approach to assessing patients with chronic pain.

Openness to experience, general intelligence and crystallized intelligence: A methodological extension

Using a semi-partial correlation approach, Gignac, Stough, and Loukomitis [Gignac, G. E., Stough, C., & Loukomitis, S. (2004). Openness, intelligence, and self-report intelligence. Intelligence, 32, 133–143] examined the relationship between Openness and ‘ g’ and residualized scores from Vocabulary and Information as estimates of crystallized intelligence. To help overcome any ambiguity in the interpretation of the results, a more sophisticated analysis based on nested factor modeling (SEM) is demonstrated to support the findings reported in Gignac et al. Researchers are encouraged to make use of the nested factor modeling approach, rather than the procedure endorsed in Gignac et al., because of advantages relevant to modeling superior estimates of intelligence factors, as well the means of evaluating model fit afforded by SEM.

Naming in Patients With Alzheimer’s Disease: Influence of Age of Acquisition and Categorical Effects

The role of age of acquisition (AoA) and other variables classically supposed to influence lexical semantic tasks is explored in Alzheimer’s disease (AD) patients. A naming test that included living and nonliving items was given to patients and controls. Measures of AoA of the test items were obtained from normal subjects. Living items were acquired earlier than nonliving items. Semipartial correlation analyses were performed to determine the independent contribution of each variable to naming. The “category” (living vs. nonliving items) was included as an independent factor. It emerged that AoA, name agreement and category (with living category predicting lower scores) were the main predictors of naming in AD patients. Only factor agreement reached significance in control groups. The hypothesis is discussed that the category dissociation may be produced by the different nature of the semantic correlation network that makes the categories differentially demanding of processing resources.

A computer simulation of the prevention of the transmission of Toxoplasma gondii on swine farms using a feline T. gondii vaccine

Transmission of Toxoplasma gondii on swine farms was investigated using a deterministic dynamic computer simulation model. A primary focus was to evaluate a feline T. gondii vaccine. Animal populations (swine and cats) were compartmentalized based on the stage of T. gondii infection. Simulations were run under conditions of closed and equilibrium population size. Model parameters were varied in a factorial experimental design to test the following hypotheses: T. gondii infection in finishing pigs decreases with (1) vaccination of susceptible cats, (2) an increase in the proportion of cats captured for vaccination, (3) a decrease in the initial number of cats, (4) a decrease in the initial T. gondii prevalence in cats and (5) a decrease in oocyst-survival time. Seeding conditions included a total of 10, 20, 30, 40 or 50 cats, initial T. gondii prevalences in cats of 30, 60 or 90%, vaccination of 0, 50 or 75% of the cats and two vaccination schedules (the field schedule from a prior trial and a weaning-vaccination schedule). Simulations were run at oocyst-survival times of 52, 39 and 26 weeks. T. gondii prevalence in finishing pigs was recorded every week for 10 years. The probability of elimination of T. gondii from finishing pigs increased with a decrease in the number of cats and a decrease in oocyst-survival time. The last-year average prevalence was used as the outcome in a multiple linear regression analysis. Decreased T. gondii prevalence in finishing pigs was the result of a decrease in the initial number of cats on the farm (squared semipartial correlation coefficient (sr 2)=47%), decreased oocyst survival (sr 2=35%), using the weaning-vaccination schedule (sr 2=7%) and vaccination versus non-vaccination (sr 2=5%). Unexpectedly, the initial T. gondii prevalence in cats had no effect on T. gondii prevalence in finishing pigs. The simulation supports the field trial indicating vaccine effectiveness. However, vaccination had less impact on decreasing T. gondii infection in finishing pigs than a decrease in the number of farm cats.

The Concurrent Validity of the Devereux Scales of Mental Disorders

The present study was designed as an initial investigation of the concurrent validity of the Devereux Scales of Mental Disorders (DSMD; Naglieri, LeBuffe, &amp; Pfeiffer, 1994) in relation to the Behavior Assessment System for Children (BASC) and the Child Behavior Checklist (CBCL) in two clinically referred samples of children and adolescents. The DSMD is a recently published multidimensional behavior rating scale comprised of empirically and conceptually derived scales designed to aid the diagnostic decision-making process for youth with severe psychiatric and educational classifications. To date, no published study has examined the concurrent validity of the DSMD. The first of the present studies compared the DSMD and BASC with 64 children and adolescents in an inpatient hospital setting; the second study compared the DSMD and CBCL with 74 children and adolescents classified as having serious emotional disturbance (SED) in residential care. Concurrent validity was assessed through correlations and squared semipartial correlation coefficients between the DSMD scale scores and the BASC and CBCL scale scores. Paired t tests (p &lt; .001) were also computed across scales of similar constructs. In comparison to the BASC and CBCL, the DSMD demonstrated strong concurrent validity between conceptually similar scales, providing some evidence for the concurrent validity of the DSMD. Specifically, relative to the other measures, the DSMD appears to have some unique scales designed to detect more acute or serious pathology, whereas the BASC and CBCL may be better equipped to assess specific types of externalizing symptoms.

Sample Size Requirements for Accurate Estimation of Squared Semi-Partial Correlation Coefficients

The increase in the squared multiple correlation coefficient associated with a variable in a regression equation is a commonly used measure of importance in regression analysis. An alternative measure of importance is the difference in the adjusted squared multiple correlation coefficients. Both estimate the difference in the population squared multiple correlation coefficients (Δρ2), a quantity also called a squared semi-partial correlation coefficient. In planning a study that will use regression analysis, it is important to select a sample size that will allow Δρ2 to be estimated with adequate accuracy. Results showed that the sample size necessary for adequate accuracy depend strongly on three factors: (a) the population squared multiple correlation coefficient (ρ2), (b) the population increase in ρ2, and (c) the desired degree of accuracy. The number of predictors had a small effect on the required sample size. Tables to facilitate sample size selection were presented.

The impact of a clinic-based literacy intervention on language development in inner-city preschool children.

To determine the effect of a clinic-based literacy intervention on the language development of preschool children. A convenience sample of families presenting to 2 urban pediatric clinics for well-child care met the following criteria: the family was Latino or black and English- or Spanish-speaking; the child was 2 to 5.9 years old, with no neurodevelopmental disability, at a gestational age of 34 weeks or more, and not attending kindergarten. Participants at the first clinic (intervention group) were exposed to a literacy support program, based on Reach Out and Read (ROR), during the previous 3 years. At the second clinic (comparison group), a similar program started 3 months before the study. Parent-child reading activities were measured using the READ Subscale of the StimQ. Language development was measured using the One-Word Expressive and Receptive Picture Vocabulary Tests, and was performed in the child's primary language. A total of 122 study participants (49 interventions and 73 comparisons) met inclusion criteria and completed all measures. Intervention and comparison families were similar for most sociodemographic variables. Intervention families reported reading together with their children approximately 1 more day per week. Intensity of exposure to ROR (measured by total number of contacts with the program) was associated with increased parent-child reading activities, as measured by the StimQ-Read Subscale (r = 0.20). Intervention children had higher receptive language (mean: 94.5 vs 84.8) and expressive language (mean: 84.3 vs 81.6). After adjusting for potential confounders in a multiple regression analysis, intervention status was associated with an 8.6-point increase (95% confidence interval [CI]: 3.3, 14.0) in receptive language (semipartial correlation [SR]coefficient = 0.27), and a 4.3-point increase (95% CI: 0.04, 8.6) in expressive language (SR = 0.17). In a similar multiple regression, each contact with ROR was associated with an adjusted mean 0.4-point increase (95% CI: 0.1, 0.6) in receptive score, and an adjusted mean 0.21-point increase (95% CI: 0. 02, 0.4) in expressive score. ROR is an important intervention, promoting parental literacy support and enhancing language development in impoverished preschool children. Integration of literacy promoting interventions such as these into routine pediatric health care for underserved populations can be recommended.

The influence of marital adjustment on 3-year left ventricular mass and ambulatory blood pressure in mild hypertension.

Of psychosocial stressors, job strain has been associated with a sustained increase in blood pressure. The impact of marital factors on blood pressure and target organ has not been explored. To evaluate whether marital adjustment, measured at baseline by self-report (Dyadic Adjustment Scale) influences left ventricular mass index (LVMI) and ambulatory blood pressure measured over 3 years in patients with mild hypertension. A prospective cohort study was conducted on 103 cohabiting males or females, including 72 with technically adequate echocardiograms, who at baseline were unmedicated, employed, and living with a significant other, all for a minimum of 6 months and had repeated elevated office diastolic blood pressure. Left ventricular mass by M-mode echocardiography indexed to body surface area and blood pressure were measured by ambulatory blood pressure every 15 minutes (daytime) and hourly between 11 PM and 7 AM. Marital adjustment, smoking, drinking, and baseline LVMI contributed significantly to the prediction of 3-year LVMI (semipartial correlation, sr(2) = 0.04, 0.07, 0.03, and 0.22; P =.03,.008,.08, and <.001, respectively) together accounting for 36% of the total variability in follow-up LVMI. Three-year ambulatory blood pressure measures were not significantly related to marital adjustment but there were correlations with Dyadic Adjustment Scale subscales. Low or high levels of spousal contact during 3-year ambulatory blood pressure monitoring were associated with an increase or decrease of 3-year, 24-hour diastolic blood pressure, consistent with the quality of marital adjustment (P =.04) or marital satisfaction (Dyadic Adjustment Scale subscale, P =.008). In a cohort of subjects with mild essential hypertension, marital adjustment had an influence on 3-year LVMI. Depending on the quality of marital adjustment, spousal contact at 3 years was associated with an increase or decrease of 3-year diastolic blood pressure. Confirmation of these results, including objective marital assessment and the participation of normotensive subjects, is required. Arch Intern Med. 2000;160:3453-3458.

Knowledge of Diabetes mellitus and Glycaemic Control

It has been claimed that lack of knowledge of diabetes mellitus has been a cause of excess admissions and morbidity. There is also some evidence that diabetic education programs can improve self-regulatory behaviour and reduce hospital admissions. In this report we test the hypothesis that greater knowledge of diabetes should be associated with better glycaemic control and lower fasting blood glucose levels. Responses to diabetes knowledge questions were provided by 420 patients attending diabetic clinics in Kuwait. In an earlier and complementary report a principal components analysis revealed that knowledge of diabetes cannot be understood in terms of a single general factor. With a subsequent Varimax rotation we obtained 12 uncorrelated knowledge factors with eigenvalues greater than unity and these would all be confounded in the simple aggregation of correct answers to diabetes knowledge questions employed in earlier studies. Results are now described which show that the 12-factor model of diabetes knowledge is better able to predict blood glucose levels than the scores obtained on a single scale by just summing the correct answers to all diabetes knowledge questions. A standard multiple linear regression with the diabetes knowledge factors age, sex and ‘years since diagnosis’ as independent variables, and fasting blood glucose levels as the dependent variable, shows that 4 of the 12 factors yield statistically significant semi-partial correlation coefficients that account for unique fractions of the total variance of blood glucose levels. The meaning of these 4 factors is discussed with special reference to glycaemic control and blood glucose levels.

Link between Orthogonal and Standard Multiple Linear Regression Models

Several topics in connection with a recently proposed method for the orthogonalization of predictor variables (dominant component analysis) are considered. Applying the sequential regression procedure, it is shown that dominant component analysis and the standard multiple linear regression method are directly related to each other. In addition, it is demonstrated that an earlier proposed iterative procedure for the orthogonalization of a correlated variable can be efficiently replaced by one step regression. It is also shown that the coefficient of determination for an orthogonal descriptor coincides with the corresponding squared semipartial correlation coefficient. Finally, the origin of extra information in an orthogonalized predictor variable is discussed.

Higher plasma IGF-1 levels are associated with increased delta sleep in healthy older men.

Sleep quality declines with age, with less time in deep or slow wave sleep (SWS) and reduced amplitude of the delta waves that characterize it. Age-related declines also occur in lean body mass, growth hormone (GH), and insulin-like growth factor 1 (IGF-1). These changes in sleep quality and anabolic status may be related, as administration of GH or growth hormone releasing hormone (GHRH) can enhance SWS and decrease awakenings in young men. Here we examine the relationship between plasma IGF levels and delta sleep quality in older men. The sleep EEG of 30 healthy elderly men (64 +/- 6 yrs; range 50-75) was recorded on the second of 2 consecutive nights. Plasma samples were drawn within 3 weeks of EEG recording, and IGF levels were assayed by RIA after acid extraction. IGF explained 28% (semi-partial correlation coefficient r = .53; p = .003) of the variance in average delta energy per epoch of SWS, after age-related variance was removed. Higher IGF was associated with higher average delta energy. Similar results were obtained for total delta energy during SWS (r = .37, p = .04) 4nd time spent in SWS (r = .42, p = .02). Other measures of sleep quality (e.g., wakefulness, REM sleep) were not correlated with IGF. The IGF delta relationship was minimally influenced by moderator variables such as thyroxine (T3, T4), and/or body mass index (BMI). We conclude that age-adjusted IGF levels in healthy senior men co-vary significantly with SWS and the delta energy that characterizes it.

Suicidal ideation in high school students: depression and other correlates.

English-Canadian high school students (129 boys, 117 girls) participated in a study of the relationship between suicidal ideation and selected personal variables, stress, and social support. Associations were found between suicidal ideation and the variables of gender, self-esteem, locus of control, depression, drug use, stress, perception of health, family status, academic performance, social support, and anomie. Multiple regression analysis identified depression and alcohol use as best individual predictors among these variables. Semi-partial correlation analyses showed that removal of the effect of depression resulted in a loss of initially significant relationships between suicidal ideation and the other variables except for alcohol use, drug use, and health satisfaction.