Effect of the violation of assumptions upon significance levels of the Pearson r.

Conducted a study to determine empirically the effects of the violation of assumptions of normality and of measurement scales on the Pearson product-moment correlation coefficient. The effects of such violations were studied separately and in combination for samples varying in size from 5 to 60. Monte Carlo procedures were used to generate populations of scores for 4 basic distributions: normal, positively skewed, negatively skewed, and leptokurtic. Samples of varying sizes were then randomly selected from specific populations. Results were based on distributions of r , which were calculated on 5,000 sets of samples of n = 5 or n = 15 and 3,000 sets of samples of n = 30 or n = 60. Results indicate that the Pearson r is insensitive to extreme violations of the basic assumptions of normality and of the type of measurement scale. Failure to meet the basic assumptions separately or in combination had little effect on the obtained distributions of r s based on rho of zero. (18 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

An Empirical Evaluation of Alternative Methods of Estimation for Confirmatory Factor Analysis With Ordinal Data.

Confirmatory factor analysis (CFA) is widely used for examining hypothesized relations among ordinal variables (e.g., Likert-type items). A theoretically appropriate method fits the CFA model to polychoric correlations using either weighted least squares (WLS) or robust WLS. Importantly, this approach assumes that a continuous, normal latent process determines each observed variable. The extent to which violations of this assumption undermine CFA estimation is not well-known. In this article, the authors empirically study this issue using a computer simulation study. The results suggest that estimation of polychoric correlations is robust to modest violations of underlying normality. Further, WLS performed adequately only at the largest sample size but led to substantial estimation difficulties with smaller samples. Finally, robust WLS performed well across all conditions. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Independent effects of Alzheimer's disease on neuropsychological functioning.

A new analytical procedure, single common factor analysis, was carried out on the data from a relatively large sample of normals (n?=?101) and patients with Alzheimer's disease (AD; n?=?180) to examine the extent to which there were independent effects of disease status on different neuropsychological variables. This technique uses structural equation methods to determine what all of the variables have in common, and then controls this common factor when examining the relationship between diagnostic group and each individual test variable. To the extent that AD represents the sum of independent breakdowns of different information processing domains, then there should be sets of variables that have weak or nonexistent links to the other variables. However, the results revealed that a large proportion of the AD-related effects on test scores was shared and was not independent of the AD-related effects on other variables. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Toward a theory of hypnotic behavior: Effects on suggestibility of defining the situation as hypnosis and defining response to suggestions as easy.

96 Ss participated in a 2 X 2 factorial experiment which was designed to assess the effects on suggestibility of: defining the situation as hypnosis or as control, and defining response to suggestions as easy or as difficult. The dependent variables consisted of responses to 8 standardized test suggestions (Barber Suggestibility Scale). Ss told "You are in the hypnosis group" were more responsive to suggestions than Ss told "You are in the control group." Ss told that it was easy to respond to test suggestions were more suggestible than Ss told that it was difficult. The suggestibility-enhancing effects of the independent variables were additive: the level of suggestibility was highest when the situation was defined as hypnosis and the test suggestions as easy; next highest when either the situation was defined as hypnosis or the suggestions as easy; and lowest when the situation was defined as control and the suggestions as difficult. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Power of the independent samples t test under a prevalent psychometric measure distribution.

Studies in psychology often have low power because of inadequate sample size. Thus, recent articles in the Journal of Consulting and Clinical Psychology have suggested making sample size determinations through readily available tables that are based on population normality. Questions have been raised on the use of these power tables because prevalent psychometric distributions, such as the discrete mass at zero with gap that occurs with 1st use or onset variables, are radically nonnormal. In addition to demonstrating the robustness of the independent samples t test with respect to Type I error, the major finding of this study shows that researchers may use these power tables without modification for this radically nonnormal distribution. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Parametric statistics and levels of measurement: Factorial designs and multiple regression.

Let Y be a continuous, ordinal measure of a latent variable Θ. In general, for factorial designs, an analysis of variance of the observed variable Y cannot be used to draw inferences about main effects and interactions on the latent variable Θ even when the standard normality and equality of variance assumptions hold. If Y is a continuous, ordinal measure of a latent variable Θ; X?,…, Xn are continuous, ordinal measures of latent variables Φ?,…, Φn; and the observed measures have a multivariate normal distribution, then a multiple regression analysis of the observed criterion measure Y and predictors X?,…, Xn can be used to test hypotheses about multivariate associations among the latent variables. Furthermore, the predicted values Y′ are unbiased estimates of quantities that are monotonically related to predicted values on the latent criterion variable Θ. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Factors that influence the temporal stability of personality by questionnaire.

Analyzed test–retest reliability data gathered from 106 sources (89 independent samples), using a multiple-regression method in an attempt to estimate the effects of several factors on questionnaire stability. We examined 8 self-report inventories: the High School Personality Questionnaire, the 16PF, the MMPI, the Myers–Briggs Type Indicator, the CPI, the Guilford–Zimmerman Temperament Survey, the EPPS, and the OPI. Samples ranged in size and encompassed a wide range of Ss divergent on status and age. We found S's age and status, number of test items, test interitem correlation, and test–retest interval to be significant predictors of reliability. Variables representing general adjustment were found to be less predictable than extraversion variables, and short-term reliability was more predictable than long-term reliability. S's sex and specific questionnaires were not found to have a significant effect on reliability. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Heterozygosity and retention of ejaculatory reflex after castration in male mice.

10 genotypes were used as quantified independent variables to test the hypothesis that degree of heterozygosity is associated with retention of the ejaculatory reflex after castration in 156 male mice. The hypothesis was supported when a significant positive correlation was found between percent heterozygosity and days to last postcastrational ejaculatory reflex. However, results also indicate that unknown environmental factors contribute importantly to the variance. Findings are discussed in terms of environmental and intra- and interspecific genetic variables affecting the trait. (21 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Variations in the administration of the Self-Directed Search, scoring accuracy, and satisfaction with results.

Examined the effects of the test administrator's attitudes toward Holland's Self-Directed Search, the size of the group taking the instrument, and whether monitors were used during the administration on both self-scoring accuracy and satisfaction with results on the Self-Directed Search. Over 25% of the 184 college freshmen who took the instrument during orientation made scoring errors resulting in incorrect high-point codes, and over 50% obtained incorrect summary codes. Of the 3 independent variables, only monitoring reduced self-scoring errors, and none affected satisfaction. Questions are raised about whether, even with monitoring, error rates are too high and satisfaction too low to warrant the use of the instrument as a self-counseling device. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A causal model of the relationship among accidents, biodata, personality, and cognitive factors.

The purpose of this research was to construct and test a causal model of the accident process. Data were gathered on 362 chemical industry workers. The causal model was analyzed and cross-validated using LISREL VI. It was proposed that social maladjustment traits, some characteristics of neurosis, cognitive ability, employee age, and job experience would have independent causal effects on the accident criterion, even when the effects of accident risk and involvement in counseling were controlled. Two rationally derived, content-validated scales based on MMPI items were created to measure social maladjustment and the aspects of neurosis that result in a state of distractibility. The results showed the causal model as a whole to be viable in the initial and cross-validation analyses, and the social maladjustment and distractibility variables were found to be significant causal parameters of accidents. This study developed a new direction for future accident research by its use of causal modeling and by the creation of two new scales for the assessment of employee accident potential. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Canonical correlation analysis: A general parametric significance-testing system.

Suggests that significance tests for 9 of the most common statistical procedures (simple correlation, t test for independent samples, multiple regression analysis, 1-way ANOVA, factorial ANOVA, analysis of covariance, t test for correlated samples, discriminant analysis, and chi-square test of independence) can all be treated as special cases of the test of the null hypothesis in canonical correlation analysis for 2 sets of variables. (15 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Decreased stature in gestational diabetes mellitus

Short stature has been associated with various degrees of abnormal glucose tolerance in middle-aged people, where the effects of age and metabolic control would be difficult to exclude. We chose to examine body stature in women with gestational diabetes mellitus (GDM), a prediabetic state affecting a young group of people. A sample of 2772 Greek pregnant women, referred for GDM screening was examined. After a 100-g oral glucose tolerance test, 1787 women were classified as normal (N), 300 women were found with one abnormal glucose value (OAV) and 685 women with GDM. Basal insulin resistance was calculated in 640 women by homeostasis model assessment. In addition, 51 pregnant women with pre-existing Type II (non-insulin-dependent) diabetes mellitus and 109 with pre-existing Type I (insulin-dependent) diabetes mellitus were included in the study. There was a gradual decrease in mean height (cm) as glucose intolerance became more severe: N: 161.0 +/- 6.2, OAV:160.2 +/- 6.1, GDM:158.7 +/- 6.3, Type II diabetes 158.2 +/- 7.0 (p < 0.001, analysis of variance]. Height in Type I diabetes (160.1 +/- 5.9) did not differ from the normal group. The difference in height between the normal and GDM groups remained (p < 0.001) when body weight, age, birth before or after 1960 and educational status were also taken into account. An independent correlation was also found between height and insulin resistance (n = 640) adjusted for the above mentioned variables. In conclusion, short stature appears to be associated with glucose intolerance as an independent variable, even when this intolerance is both mild and temporary. The previously unrecognised independent association of stature with basal insulin resistance merits further investigation.     

Nonlinearity of range corrections in meta-analysis: Test of an improved procedure.

Evaluated an improved procedure for range-restriction corrections in meta-analysis. When population correlations were approximately normally distributed, the new nonlinear range correction procedure improved the accuracy of the Schmidt-Hunter (S-H) interactive method in estimating both the mean (Mρ) and standard deviation (SDρ) of population correlations, making it the most accurate of the procedures examined. In the homogeneous case (SDρ?=?0), the nonlinear range correction again improves accuracy of estimates of SDρ. In this important case, Taylor Series Approximation (TSA) 1 is considerably less accurate than the S-H interactive and the TSA 2 procedures, and the N. S. Raju et al (see record 1991-26294-001) procedure yields the least accurate estimates of SDρ. Finally, it was found that the nonlinear range-correction procedure produces improvements in the accuracy of the interactive method even under extreme violations of the normality assumption. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A common language effect size statistic.

Some of the shortcomings in interpretability and generalizability of the effect size statistics currently available to researchers can be overcome by a statistic that expresses how often a score sampled from one distribution will be greater than a score sampled from another distribution. The statistic, the common language effect size indicator, is easily calculated from sample means and variances (or from proportions in the case of nominal-level data). It can be used for expressing the effect observed in both independent and related sample designs and in both 2-group and n-group designs. Empirical tests show it to be robust to violations of the normality assumption, particularly when the variances in the 2 parent distributions are equal. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Distinguishing between moderator and quadratic effects in multiple regression.

Moderated regression analysis is commonly used to test for multiplicative influences of independent variables in regression models. D. Lubinski and L. G. Humphreys (1990) have shown that significant moderator effects can exist even when stronger quadratic effects are present. They recommend comparing effect sizes associated with both effect types and selecting the model that yields the strongest effect. The authors show that this procedure of comparing effect sizes is biased in favor of the moderated model when multicollinearity is high because of the differential reliability of the quadratic and multiplicative terms in the regression models. Fortunately, levels of multicollinearity under which this bias is most problematic may be outside the range encountered in many empirical studies. The authors discuss causes and implications of this phenomenon as well as alternative procedures for evaluating structural relationships among variables. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Investigations of absenteeism: Using event history models to study the absence-taking process.

Traditional correlational and regression analyses of aggregated absences are ill-suited for estimating parameters of the absence-taking process. A class of event history models is described that effectively deals with the distributional and temporal properties of absences. One such model, a proportional hazard rate model, was applied to the daily attendance records of 2,130 incumbent, white-collar employees of a national financial service organization. Temporal and historical variables affected the hazard rate of voluntary absence taking. Demographic variables that were significantly correlated with aggregated absences did not improve prediction of the hazard rate. Consequences of violating the modeling assumptions of independent observations and large samples were examined empirically. Results indicated that the model was robust to such violations. Implications of the findings and of the successful application of the model are discussed. Several recommendations are made for using the model in organizations and for future studies of absence taking and other low base-rate events in organizations. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

The influence of nicotine pretreatment on mesoaccumbens dopamine overflow and locomotor responses to D-amphetamine

In this retrospective cephalometric study of mandibular rotational change, we studied 60 cases. All patients were treated with full fixed appliances and exhibited an opening rotation of at least 1.5 degrees during treatment, as measured on the basis of the angle of the Y-axis to the sella-nasion line (SN). Our objective was to determine whether this opening rotation was sustained during retention. Paired t tests were used to test the hypothesis that the treatment change or opening rotation was stable and that these patients did not return to their original mandible-to-cranium relationship in the posttreatment period. Stepwise regression analysis was used to determine which (if any) changes in the independent variables during treatment could predict the subsequent behavior of the angle of the mandibular plane to the SN and the angle of the Y-axis to the SN during retention. During treatment, the mean increase in the angle of the Y-axis to the SN was 2.43 degrees. After an average posttreatment period of 54 months, this angle was reduced on average by only 0.73 degrees. Stepwise linear-regression analysis indicated that none of the treatment changes seen in the independent variables strongly predicted the ensuing closing rotation seen during retention. The correlation coefficient between the Y-axis angle and the mandibular-plane angle during treatment was 0.67. Mandibular opening rotations as a consequence of orthodontic treatment do not invariably return to the pretreatment value, and their negative effects--although sometimes small--cannot be discounted. Because the preponderant evidence of a closing rotation occurs in the terminal pubertal growth stages, the net effect may be even more significant.     

Removing the Shadow of Suspicion: The Effects of Apology Versus Denial for Repairing Competence- Versus Integrity-Based Trust Violations.

Two studies were conducted to examine the implications of an apology versus a denial for repairing trust after an alleged violation. Results reveal that trust was repaired more successfully when mistrusted parties (a) apologized for violations concerning matters of competence but denied culpability for violations concerning matters of integrity, and (b) had apologized for violations when there was subsequent evidence of guilt but had denied culpability for violations when there was subsequent evidence of innocence. Supplementary analyses also revealed that the interactive effects of violation type and violation response on participants' trusting intentions were mediated by their trusting beliefs. Combined, these findings provide needed insight and supporting evidence concerning how trust might be repaired in the aftermath of a violation. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

To predict some of the people some of the time: A reexamination of the moderator variable approach in personality theory.

Bem and Allen (1974) proposed but did not test the notion that trait relevance moderates cross-situational behavior consistency, but did obtain support, albeit weaker than is generally recognized, for a moderator effect of self-reported consistency. Kenrick and Stringfield (1980) showed that self-reported observability of behavior may also moderate self–peer consistency, but replication attempts were not entirely successful. Our investigation produced significant moderator effects on self–peer correlations for trait relevance, consistency, and observability when each of these variables was analyzed alone; we also obtained significant moderator effects for trait relevance and self-reported consistency (but not for observability) when all the variables were included in the same regression analysis. There was significant support for the prediction that the moderators combine additively: Self–peer correlations increased linearly as a function of the number of contributing moderator effects. There was even stronger support for the prediction that the moderators combined in an all-or-nothing manner: Self–peer correlations were high when the level of all three moderators was high, and low when the level of one or more of the three moderators was low. Our future research recommendation? Identify the optimal moderators and the manner in which they can be combined. (PsycINFO Database Record (c) 2010 APA, all rights reserved)