The effects of clustered data on standard error estimates in covariance structure analysis : a field data application

Purpose – The purpose of this article is to present an empirical analysis of complex sample data with regard to the biasing effect of non-independence of observations on standard error parameter estimates. Using field data structured in the form of repeated measurements it is to be shown, in a two-factor confirmatory factor analysis model, how the bias in SE can be derived when the non-independence is ignored.

Design/methodology/approach – Three estimation procedures are compared: normal asymptotic theory (maximum likelihood); non-parametric standard error estimation (naïve bootstrap); and sandwich (robust covariance matrix) estimation (pseudo-maximum likelihood).

Findings – The study reveals that, when using either normal asymptotic theory or non-parametric standard error estimation, the SE bias produced by the non-independence of observations can be noteworthy.

Research limitations/implications – Considering the methodological constraints in employing field data, the three analyses examined must be interpreted independently and as a result taxonomic generalisations are limited. However, the study still provides “case study” evidence suggesting the existence of the relationship between non-independence of observations and standard error bias estimates.

Originality/value – Given the increasing popularity of structural equation models in the social sciences and in particular in the marketing discipline, the paper provides a theoretical and practical insight into how to treat repeated measures and clustered data in general, adding to previous methodological research. Some conclusions and suggestions for researchers who make use of partial least squares modelling are also drawn.

Robust filtering for uncertain discrete-time systems with uncertain noise covariance and uncertain observations

The use of Kalman filtering is very common in state estimation problems. The problem with Kalman filters is that they require full prior knowledge about the system modeling. It is also assumed that all the observations are fully received. In real applications, the previous assumptions are not true all the time. It is hard to obtain the exact system model and the observations may be lost due to communication problems. In this paper, we consider the design of a robust Kalman filter for systems subject to uncertainties in the state and white noise covariances. The systems under consideration suffer from random interruptions in the measurements process. An upper bound for the estimation error covariance is proposed. The proposed upper bound is further minimized by selection of optimal filter parameters. Simulation example shows the effectiveness of the proposed filter.

Variance-covariance based weighing for neural network ensembles

Neural network (NN) is a popular artificial intelligence technique for solving complicated problems due to their inherent capabilities. However generalization in NN can be harmed by a number of factors including parameter's initialization, inappropriate network topology and setting parameters of the training process itself. Forecast combinations of NN models have the potential for improved generalization and lower training time. A weighted averaging based on Variance-Covariance method that assigns greater weight to the forecasts producing lower error, instead of equal weights is practiced in this paper. While implementing the method, combination of forecasts is done with all candidate models in one experiment and with the best selected models in another experiment. It is observed during the empirical analysis that forecasting accuracy is improved by combining the best individual NN models. Another finding of this study is that reducing the number of NN models increases the diversity and, hence, accuracy.

Habituation of male sexual arousal: effects of attentional focus

Three experiments are reported demonstrating that levels of penile tumescence and subjective sexual arousal are greater when men employ participant-oriented rather than spectator-oriented attentional focus while viewing an erotic film segment. Under each instructional set, there was a reduction in sexual arousal during repeated erotic stimulation. As sexual arousal habituated, the men reported feeling less absorbed during erotic stimulation. When these associated changes in attentional focus (absorption) were partialled out through analysis of covariance, sexual arousal remained relatively stable over trials, suggesting that sexual arousal is less likely to habituate if attentional focus remains constant during repeated erotic stimulation. Further directions for studying associations between habituation of sexual arousal and cognitive processing are discussed.

Regional specificity of exercise and calcium during skeletal growth in girls: a randomized controlled trial

Combining exercise with calcium supplementation may produce additive or multiplicative effects at loaded sites; thus, we conducted a single blind, prospective, randomized controlled study in pre- and early-pubertal girls to test the following hypotheses. (1) At the loaded sites, exercise and calcium will produce greater benefits than exercise or calcium alone. (2) At non-loaded sites, exercise will have no benefit, whereas calcium with or without exercise will increase bone mass over that in exercise alone or no intervention. Sixty-six girls aged 8.8 ± 0.1 years were randomly assigned to one of four study groups: moderate-impact exercise with or without calcium or low-impact exercise with or without calcium. All participants exercised for 20 minutes, three times a week and received Ca-fortified (434 ± 19 mg/day) or non-fortified foods for 8.5 months. Analysis of covariance (ANCOVA) was used to determine interaction and main effects for exercise and calcium on bone mass after adjusting for baseline bone mineral content and growth in limb lengths. An exercise-calcium interaction was detected at the femur (7.1%, p < 0.05). In contrast, there was no exercise-calcium interaction detected at the tibia-fibula; however, there was a main effect of exercise: bone mineral content increased 3% more in the exercise than non-exercise groups (p < 0.05). Bone mineral content increased 2-4% more in the calcium-supplemented groups than the non-supplemented groups at the humerus (12.0% vs. 9.8%, respectively, p < 0.09) and radius-ulna (12.6% vs. 8.6%, respectively, p < 0.01). In conclusion, greater gains in bone mass at loaded sites may be achieved when short bouts of moderate exercise are combined with increased dietary calcium, the former conferring region-specific effects and the latter producing generalized effects.

Exercise and calcium combined results in a Greater osteogenic effect than either factor alone: a blinded randomized placebo-controlled trial in boys

We examined the combined effects of exercise and calcium on BMC accrual in pre- and early-pubertal boys. Exercise and calcium together resulted in a 2% greater increase in femur BMC than either factor alone and a 3% greater increase in BMC at the tibia–fibula compared with the placebo group. Increasing dietary calcium seems to be important for optimizing the osteogenic effects of exercise.

Introduction: Understanding the relationship between exercise and calcium during growth is important given that the greatest benefits derived from these factors are achieved during the first two decades of life. We conducted a blinded randomized-controlled exercise–calcium intervention in pre- and early-pubertal boys to test the following hypotheses. (1) At the loaded sites (femur and tibia–fibula), exercise and calcium will produce greater skeletal benefits than either exercise or calcium alone. (2) At nonloaded sites (humerus and radius–ulna), there will be an effect of calcium supplementation.

Materials and Methods: Eighty-eight pre- and early-pubertal boys were randomly assigned to one of four study groups: moderate impact exercise with or without calcium (Ca) (Ex + Ca and Ex + placebo, respectively) or low impact exercise with or without Ca (No-Ex + Ca and No-Ex + Placebo, respectively). The intervention involved 20 minutes of either moderate- or low-impact exercise performed three times a week and/or the addition of Ca-fortified foods using milk minerals (392 ± 29 mg/day) or nonfortified foods over 8.5 months. Analysis of covariance was used to determine the main and combined effects of exercise and calcium on BMC after adjusting for baseline BMC.

Results: At baseline, no differences were reported between the groups for height, weight, BMC, or bone length. The increase in femur BMC in the Ex + Ca group was 2% greater than the increase in the Ex + placebo, No-Ex + Ca, or No-Ex + Placebo groups (all p < 0.03). At the tibia–fibula, the increase in BMC in the Ex + Ca group was 3% greater than the No-Ex + placebo group (p < 0.02) and 2% greater than the Ex + Placebo and the No-Ex + Ca groups (not significant). No effect of any group was detected at the humerus, ulna–radius, or lumbar spine for BMC, height, bone area, or volume.

Conclusions: In this group of normally active boys with adequate calcium intakes, additional exercise and calcium supplementation resulted in a 2–3% greater increase in BMC than controls at the loaded sites. These findings strengthen the evidence base for public health campaigns to address both exercise and dietary changes in children for optimizing the attainment of peak BMC.

The application of rough set and Kohonen network to feature selection for object extraction

Selecting a set of features which is optimal for a given task is a problem which plays an important role in a wide variety of contexts including pattern recognition, images understanding and machine learning. The paper describes an application of rough sets method to feature selection and reduction in texture images recognition. The proposed methods include continuous data discretization based on Kohonen neural network and maximum covariance, and rough set algorithms for feature selection and reduction. The experiments on trees extraction from aerial images show that the methods presented in this paper are practical and effective.

Optimal multisensor data fusion for linear systems with missing measurements

Multisensor data fusion has attracted a lot of research in recent years. It has been widely used in many applications especially military applications for target tracking and identification. In this paper, we will handle the multisensor data fusion problem for systems suffering from the possibility of missing measurements. We present the optimal recursive fusion filter for measurements obtained from two sensors subject to random intermittent measurements. The noise covariance in the observation process is allowed to be singular which requires the use of generalized inverse. Illustration example shows the effectiveness of the proposed filter in the measurements loss case compared to the available optimal linear fusion methods.

There's SEM and 'SEM' : a critique of the use of PLS regression in information systems research

In disciplines other than IS, the use of covariance-based structural equation modelling (SEM) is the mainstream method for SEM analysis, and for confirmatory factor analysis (CFA). Yet a body of IS literature has developed arguing that PLS regression is a superior tool for these analyses, and for establishing reliability and validity. Despite these claims, the views underlying this PLS literature are not universally shared. In this paper the authors review the PLS and mainstream SEM literatures, and describe the key differences between the two classes of tools. The paper also canvasses why PLS regression is rarely used in management, marketing, organizational behaviour, and that branch of psychology concerned with good measurement – psychometrics. The paper offers some practical options to Australasian researchers seeking greater mastery of SEM, and also acts as a roadmap for readers who want to check for themselves what the mainstream SEM literature has to say.

The Samuelson hypothesis in futures markets : an analysis using intraday data

This paper considers the Samuelson hypothesis, which argues that the futures price volatility increases as the futures contract approaches its expiration. Utilizing intraday data from 20 futures markets in six futures exchanges, we find strong support for the Samuelson hypothesis in agricultural futures. However, the Samuelson hypothesis does not hold for other futures contracts. We also provide supporting evidence that the ‘negative covariance’ hypothesis is the key factor for the empirical support of the Samuelson hypothesis. In addition, our findings remain largely unaltered even after we control for seasonality and liquidity effects.

The impact of using a pedometer on time spent walking in older adults with Type 2 diabetes

Purpose

The purpose of this study was to investigate the impact of using a pedometer on time spent walking, in sedentary and overweight adults with type 2 diabetes participating in a coaching intervention. It was hypothesized that participants using a pedometer would spend more time walking than would nonpedometer participants.

Method

A sample of 57 men and women with a mean age of 62 years participated in a randomized controlled trial in a community setting. Participants were allocated to either a pedometer and coaching (intervention) group or a coaching-only (control) group. Coaching for both groups involved education, goal setting, and supportive/ motivational strategies to increase time spent walking. The duration of the study was 6 months, with blood pressure, glycosylated hemoglobin, anthropometric, and fitness measurements assessed at baseline and at 3-month intervals.

Results

A repeated-measures analysis of variance indicated that the coaching-only group spent significantly more time walking than did the pedometer group. However, when an analysis of covariance with all the other variables as covariates was performed, group membership had no influence on time spent walking. Significant reductions in waist circumference and weight were achieved for both groups from baseline to 6 months. Cardiovascular fitness also increased significantly for both groups.

Conclusion

The study demonstrated that previously sedentary older adults with type 2 diabetes, supported with a coaching intervention, were able to achieve the physical activity targets known to be beneficial to health. However, using a pedometer added no further benefit. Further research on the impact of specific coaching strategies in diabetes management is warranted.

Variation in age and physical status prior to total knee and hip replacement surgery : a comparison of centers in Australia and Europe

Objective : To investigate whether variation exists in the preoperative age, pain, stiffness, and physical function of people undergoing total knee replacement (TKR) and total hip replacement (THR) at several centers in Australia and Europe.
Methods : Individual Western Ontario and McMaster Universities Osteoarthritis Index data (range 0-100, where 0 = best and 100 = worst) collected within 6 weeks prior to primary TKR and THR were extracted from 16 centers (n = 2,835) according to specified eligibility criteria. Analysis of covariance was used to evaluate differences in pain, stiffness, and physical function between centers, with adjustment for age and sex.
Results : There was marked variation in the age of people undergoing surgery between the centers (TKR mean age 67-73 years; F[6,1004] = 4.21, P < 0.01, and THR mean age 63-72 years; F[14,1807] = 7.27, P < 0.01). Large differences in preoperative status were observed between centers, most notably for pain (TKR adjusted mean pain 52.5-61.1; F[6,1002] = 4.26, P < 0.01, and THR adjusted mean pain 49.2-65.7; F[14,1802] = 8.44, P < 0.01) and physical function (TKR adjusted mean function 52.7-61.4; F[6,1002] = 5.27, P < 0.01, and THR adjusted mean function 53.3-71.0; F[14,1802] = 6.71, P < 0.01). Large effect sizes (up to 0.98) reflect the magnitude of variation between centers and highlight the clinical relevance of these findings.
Conclusion : The large variations in age and preoperative status indicate substantial differences in the timing of joint replacement across the centers studied, with potential for compromised surgical outcomes due to premature or delayed surgery. Possible contributing factors include patient preferences, the absence of concrete indications for surgery, and the capacity of the health care systems.

Optimal linear data fusion for systems with missing measurements

In this paper, we provide the optimal data fusion filter for linear systems suffering from possible missing measurements. The noise covariance in the observation process is allowed to be singular which requires the use of generalized inverse. The data fusion process is made on the raw data provided by two sensors  observing the same entity. Each of the sensors is losing the measurements in its own data loss rate. The data fusion filter is provided in a recursive form for ease of implementation in real-world applications.