Making causal inferences from observational data.
Explore the source record for details and available documents.
Environment↗
SEARCH · PubMed Health
Results for “statistical inference”
Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
[Tests of statistical significance in three biomedical journals: a critical review].
OBJECTIVE: To describe the use of conventional tests of statistical significance and the current trends shown by their use in three biomedical journals read in Spanish-speaking countries. METHODS: All descriptive or explanatory original articles published in the five-year period of 1996 through 2000 were reviewed in three journals: Revista Cubana de Medicina General Integral [Cuban Journal of Comprehensive General Medicine], Revista Panamericana de Salud Pública/Pan American Journal of Public Health, and Medicina Clínica [Clinical Medicine] (which is published in Spain). RESULTS: In the three journals that were reviewed various shortcomings were found in their use of hypothesis tests based on P values and in the limited use of new tools that have been suggested for use in their place: confidence intervals (CIs) and Bayesian inference. The basic findings of our research were: minimal use of CIs, as either a complement to significance tests or as the only statistical tool; mentions of a small sample size as a possible explanation for the lack of statistical significance; a predominant use of rigid alpha values; a lack of uniformity in the presentation of results; and improper reference in the research conclusions to the results of hypothesis tests. CONCLUSIONS: Our results indicate the lack of compliance by authors and editors with accepted standards for the use of tests of statistical significance. The findings also highlight that the stagnant use of these tests continues to be a common practice in the scientific literature.
Periodicals as Topic↗
X-APL: an improved family-based test of association in the presence of linkage for the X chromosome.
Family-based association methods have been developed primarily for autosomal markers. The X-linked sibling transmission/disequilibrium test (XS-TDT) and the reconstruction-combined TDT for X-chromosome markers (XRC-TDT) are the first association-based methods for testing markers on the X chromosome in family data sets. These are valid tests of association in family triads or discordant sib pairs but are not theoretically valid in multiplex families when linkage is present. Recently, XPDT and XMCPDT, modified versions of the pedigree disequilibrium test (PDT), were proposed. Like the PDT, XPDT compares genotype transmissions from parents to affected offspring or genotypes of discordant siblings; however, the XPDT can have low power if there are many missing parental genotypes. XMCPDT uses a Monte Carlo sampling approach to infer missing parental genotypes on the basis of true or estimated population allele frequencies. Although the XMCPDT was shown to be more powerful than the XPDT, variability in the statistic due to the use of an estimate of allele frequency is not properly accounted for. Here, we present a novel family-based test of association, X-APL, a modification of the test for association in the presence of linkage (APL) test. Like the APL, X-APL can use singleton or multiplex families and properly infers missing parental genotypes in linkage regions by considering identity-by-descent parameters for affected siblings. Sampling variability of parameter estimates is accounted for through a bootstrap procedure. X-APL can test individual marker loci or X-chromosome haplotypes. To allow for different penetrances in males and females, separate sex-specific tests are provided. Using simulated data, we demonstrated validity and showed that the X-APL is more powerful than alternative tests. To show its utility and to discuss interpretation in real-data analysis, we also applied the X-APL to candidate-gene data in a sample of families with Parkinson disease.
Chromosomes, Human, X↗
"Battered data"--some clinical effects of the abuse of multiple regression methods: a second look at some observations on Ro07-0582 (Misonidazole).
The paper presents an argument for the wider use of simple and standard statistical methods in the design and evaluation of both laboratory and clinical experiments with toxic radiosensitizers. It also presents an argument against the frequent practice of superposing a curve estimated from laboratory observations upon scattergrams of clinical observations to thereby assist the reader to the (often fallacious) inference that the two disparate sets of observations have enough in common to be graduated by the same curve. Such a practice achieves a substitution of reasoning by analogy for the more reliable statistical procedures of estimation and of testing a hypothesis.
Animals↗
Bayesian detection of periodic mRNA time profiles without use of training examples.
BACKGROUND: Detection of periodically expressed genes from microarray data without use of known periodic and non-periodic training examples is an important problem, e.g. for identifying genes regulated by the cell-cycle in poorly characterised organisms. Commonly the investigator is only interested in genes expressed at a particular frequency that characterizes the process under study but this frequency is seldom exactly known. Previously proposed detector designs require access to labelled training examples and do not allow systematic incorporation of diffuse prior knowledge available about the period time. RESULTS: A learning-free Bayesian detector that does not rely on labelled training examples and allows incorporation of prior knowledge about the period time is introduced. It is shown to outperform two recently proposed alternative learning-free detectors on simulated data generated with models that are different from the one used for detector design. Results from applying the detector to mRNA expression time profiles from S. cerevisiae showsthat the genes detected as periodically expressed only contain a small fraction of the cell-cycle genes inferred from mutant phenotype. For example, when the probability of false alarm was equal to 7%, only 12% of the cell-cycle genes were detected. The genes detected as periodically expressed were found to have a statistically significant overrepresentation of known cell-cycle regulated sequence motifs. One known sequence motif and 18 putative motifs, previously not associated with periodic expression, were also over represented. CONCLUSION: In comparison with recently proposed alternative learning-free detectors for periodic gene expression, Bayesian inference allows systematic incorporation of diffuse a priori knowledge about, e.g. the period time. This results in relative performance improvements due to increased robustness against errors in the underlying assumptions. Results from applying the detector to mRNA expression time profiles from S. cerevisiae include several new findings that deserve further experimental studies.
Algorithms↗
A new approach for interval estimation and hypothesis testing of a certain intraclass correlation coefficient: the generalized variable method.
We consider the problems of interval estimation and hypothesis testing for the intraclass correlation coefficient in an interrater reliability study when both raters and subjects are assumed to be randomly selected from populations of raters and subjects, respectively. We propose a novel approach for the confidence interval estimation and hypothesis testing using the concepts of generalized confidence interval (GCI) and generalized P-values. A simulation study is conducted to investigate the coverage probabilities of the GCI approach relative to the modified large sample (MLS) approach. Both methods tend to provide somewhat conservative coverage. Relative to the MLS approach, the GCI approach is closer to the correct (nominal) coverage for a two-sided interval, but farther to the correct coverage for a one-sided lower interval. Unlike the MLS approach, the GCI approach can also easily provide P-values. The fact that the GCI approach is suitable for confidence interval estimation and obtaining P-values makes the GCI approach a suitable candidate for making inference about interrater reliability.
Algorithms↗
[Statistical insight to improve reliability of pharmacological studies].
When pharmaceutical scientists describe characteristics of a drug or when they decide whether it is appropriate to initiate clinical trials to determine the drug's effects in humans, their inferences are frequently grounded in information drawn from non-clinical studies. Therefore, certain and highly objective information is required. By introducing the concept of design of experiments to control some nuisance factors and performing confirmatory studies based on sample size estimation, trustworthy information can be efficiently obtained. This paper does not demand that researchers conduct an additional confirmatory study in a series of studies conducted so far. This is a reconsideration how a series of studies should be carried forward. Statistics ought to contribute much more not only to estimation or hypothetical tests after data are collected, but also to methodology of preliminary experiments and planning of studies. Cooperation with statisticians from an early stage of the studies is all the more helpful in non-clinical studies, in which, in a sense, "perfect" experiments can be conducted more than in clinical studies.
Animals↗
Statistical estimates and clinical trials.
Statistical estimates and significance tests address distinct (but related) questions using the same data. Point estimates and confidence intervals of differences are statistical estimates that address: "How LARGE is the difference in the population of interest?" A significance test addresses the question: "How LIKELY was the difference to have occurred by chance?" Because p-values deal with the existence of a real nonzero difference between treatments but not the size of that treatment difference, they cannot be used to assess clinical (practical) significance. A confidence interval is a range of values used to infer both the size of a difference and the uncertainty of the estimate. Examples illustrate how confidence intervals help us assess both the clinical significance and the statistical significance of an observed difference. The point estimate is the outcome difference actually observed in the study sample; it is also the best single-number estimate of the unknown difference in the sampled population. Point estimates, confidence intervals, and p-values extract complementary information from study data and should all be reported for major results.
Clinical Trials as Topic↗
[Evaluation of the use of statistical techniques in original articles published in the Medicina Clínica during 3 decades (1962-1992)].
BACKGROUND: The incorrect use of statistical techniques in medical articles may seriously compromise the validity of conclusions. This finding otherwise is relatively common. METHODS: A total of 84 original articles published in Medicina Clínica between 1962 and 1992 were reviewed with the aim of assessing the use and appropriateness of statistical techniques. The use of statistics, the quality of the analyses performed, and the inaccuracy of the statistical techniques used were evaluated. We also classified the statistical techniques most commonly used throughout the study period. RESULTS: There was a marked increase in the use of statistical analyses, from 8.3% in 1962 to 83.3% in 1992. It should be noted that a substantial part of this increase has been due to the use of inferential tests, which accounted up to 70% in the sample of articles published in 1992. This finding, however, was associated with an increase in the number of incorrect analyses. The most common statistical errors included assumption of normal distribution of data (with no mention of the test used to confirm this fact), mistake between standard deviation and standard error of the mean, inadequate inferences on the basis of the sample size, inappropriate use of the Student's t test, chi-square test, nonparametric tests or multivariate analyses as well as misunderstanding of linear regression and correlation. CONCLUSIONS: High standards in scientific research have been accompanied by a significant increase in the number of clinical studies with statistical analysis of data. However, this apparently favorable situation has been associated with an increase in the number of inaccurate analyses. It has been found that sophisticated statistical tests are rarely used in articles published in Medicina Clínica.
Chi-Square Distribution↗
A critique of the use of the Kolmogorov-Smirnov (KS) statistic for the analysis of BOLD fMRI data.
The use of the Kolmogorov-Smirnov (KS) statistic for testing hypotheses regarding activation in blood oxygenation level-dependent functional MRI data is critiqued on both theoretical and empirical grounds. Theoretically, it is argued that the KS test is formally unable to support inferences of interest to most neuro-imaging studies and has reduced sensitivity compared with parametric alternatives. Empirically, false-positive rates yielded by the KS test in human data collected under the null hypothesis were significantly in excess of tabular values. These excessive false-positive rates could be explained by the presence of temporal autocorrelation. We also present evidence that the distribution of blood oxygenation level-dependent functional MRI data is only slightly nonnormal, questioning the initial impetus for the use of the KS test in this context. Finally, it is noted that parametric alternatives exist that do provide adequate control of the false-positive rate and can support inferences of interest.
Algorithms↗
A comparison of frailty and other models for bivariate survival data.
Multivariate survival data arise when each study subject may experience multiple events or when study subjects are clustered into groups. Statistical analyses of such data need to account for the intra-cluster dependence through appropriate modeling. Frailty models are the most popular for such failure time data. However, there are other approaches which model the dependence structure directly. In this article, we compare the frailty models for bivariate data with the models based on bivariate exponential and Weibull distributions. Bayesian methods provide a convenient paradigm for comparing the two sets of models we consider. Our techniques are illustrated using two examples. One simulated example demonstrates model choice methods developed in this paper and the other example, based on a practical data set of onset of blindness among patients with diabetic Retinopathy, considers Bayesian inference using different models.
Bayes Theorem↗
Detecting survival effects of socioeconomic status: problems in the use of aggregate measures.
Direct measures of SES are seldom included in medical records or large databases on disease incidence or survival, forcing researchers to infer the SES characteristics of individuals from aggregate data (e.g. census tract-level income, education, etc.). This paper assesses the degree of error that results from such inference and the impact this error may have on reported relationships between SES and survival. The authors obtained both individual and census tract-level data on 536 persons diagnosed with cancer between 1980 and 1982 and monitored their survival through 1992. Pearson correlations between individual-level and census tract-level SES variables ranged between 0.2 and 0.4. Statistically significant relationships between SES and survival were observed in the models based on individual-level but not census tract-level SES data. The authors computed the degree to which inference of individual-level from census tract-level SES reduces estimates of risk ratios across SES. It appears likely that much larger numbers of observations than have been used in published studies will be needed to better understand the relationship of SES to survival and other disease outcomes.
Age Factors↗
A preliminary evaluation of treatment outcomes at a veterans' hospital's inpatient psychiatry unit.
This article describes a preliminary effort to evaluate inpatient psychiatric services at the Carl Vinson DVA Medical Center in Dublin, Georgia. The facility annually treats a large number of veterans for a variety of psychiatric disorders. To determine whether these veterans improved following care, a simple pretest-posttest group design was employed, using the SCL-90-R, to assess psychiatric symptomatology before and after inpatient treatment. Both statistically significant and practically meaningful improvements in symptomatology were evident at discharge. While the research design does not permit causal inferences, low-cost evaluations such as this one simply demonstrating that patients get better are important first steps in empirically determining the efficacy of inpatient psychiatric services, and represent one means of demonstrating accountable practice.
Adult↗
Semi-parametric inferences for association with semi-competing risks data.
In many biomedical studies, it is of interest to assess dependence between bivariate failure time data. We focus here on a special type of such data, referred to as semi-competing risks data. In this article, we develop methods for making inferences regarding dependence of semi-competing risks data across strata of a discrete covariate Z. A class of rank statistics for testing constancy of association across strata are proposed; its asymptotic properties are also derived. We develop a novel re-sampling-based technique for calculating the variances of the proposed test statistics. In addition, we develop methods for combining test statistics for assessing marginal effects of Z on the dependent censoring variable as well as its effects on association. The finite-sample properties of the proposed methodology are assessed using simulation studies, and they are applied to data from a leukaemia transplantation study.
Bone Marrow Transplantation↗
The growth of soybean under free air [CO(2)] enrichment (FACE) stimulates photosynthesis while decreasing in vivo Rubisco capacity.
Down-regulation of light-saturated photosynthesis (A(sat)) at elevated atmospheric CO(2) concentration, [CO(2)], has been demonstrated for many C(3) species and is often associated with inability to utilize additional photosynthate and/or nitrogen limitation. In soybean, a nitrogen-fixing species, both limitations are less likely than in crops lacking an N-fixing symbiont. Prior studies have used controlled environment or field enclosures where the artificial environment can modify responses to [CO(2)]. A soybean free air [CO(2)] enrichment (FACE) facility has provided the first opportunity to analyze the effects of elevated [CO(2)] on photosynthesis under fully open-air conditions. Potential ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) carboxylation (V(c,max)) and electron transport through photosystem II (J(max)) were determined from the responses of A(sat) to intercellular [CO(2)] (C(i)) throughout two growing seasons. Mesophyll conductance to CO(2) (g(m)) was determined from the responses of A(sat) and whole chain electron transport (J) to light. Elevated [CO(2)] increased A(sat) by 15-20% even though there was a small, statistically significant, decrease in V(c,max). This differs from previous studies in that V(c,max)/J(max) decreased, inferring a shift in resource investment away from Rubisco. This raised the C(i) at which the transition from Rubisco-limited to ribulose-1,5-bisphosphate regeneration-limited photosynthesis occurred. The decrease in V(c,max) was not the result of a change in g(m), which was unchanged by elevated [CO(2)]. This first analysis of limitations to soybean photosynthesis under fully open-air conditions reveals important differences to prior studies that have used enclosures to elevate [CO(2)], most significantly a smaller response of A(sat) and an apparent shift in resources away from Rubisco relative to capacity for electron transport.
Carbon Dioxide↗
Interdependence of pulse wave variables in the human fetal aorta.
The problem of inferring the state of fetal circulation from diameter pulse waves in the descending aorta was studied with the aid of statistical methods in 12 fetuses in late uncomplicated gestation. By means of a phase-locked, echo-tracking ultrasound technique, aorta diameter-time waveforms and pulse propagation velocities were measured. In a factor analysis, three independent factors explained 75% of the total variance of variables in the waveforms. Relationships between some of the variables and biophysical events are suggested; they await confirmation from animal experiments, in order to develop a practical method for assessing fetal circulation.
Aorta, Thoracic↗
Quantification of prostate shrinkage after microwave thermotherapy: a comparison of calculated cell-kill versus 3D transrectal ultrasound planimetry.
PURPOSE: To compare prostate shrinkage after transurethral microwave thermotherapy (TUMT) with calculated cell-kill. MATERIALS AND METHODS: The calculated cell-kill from 33 males with benign prostatic hyperplasia (BPH) treated with TUMT according to the ProstaLund Feedback Treatment (PLFT) method was compared to the post-treatment prostate volume change. The prostate volume was estimated with three-dimensional transrectal ultrasound (3D-TRUS) planimetry at baseline, 3, 6, and 12 months follow-up. A paired t-test was used to test the statistical significance of differences between the cell-kill volume and the prostate volume change. Linear regression was used to infer a relationship between the cell-kill and the 3D-TRUS data. The reproducibility of the 3D-TRUS method was assessed in repeated measurements. RESULTS: The mean prostate volume at baseline (N=33) was 56.1cm(3). After 3 (N=25), 6 (N=29) and 12 months (N=23), it was 45.5 cm(3), 39.7 cm(3), and 45.1cm(3), respectively. The corresponding average cell-kill volume was 16.4 cm(3), 17.1cm(3), and 17.2 cm(3), respectively. Predicted cell-kill volume was significantly larger than prostate shrinkage at 3 (p<0.0001), 6 (p=0.0002), and 12 months (p<0.0001), and showed a strong correlation at 3 and 6 months (r=0.74, p<0.0001). Correlation at 12 months was moderate (r=0.57, p=0.0041). Examination and investigation variability both averaged 2.5%. CONCLUSIONS: Cell-kill calculations of the PLFT method are proportional to the 3D-TRUS prostate shrinkage by a factor of 0.5 and have a precision of approximately +/-10 cm(3) for 90% of the patients during the first year after treatment.
Aged↗
The "equal environments assumption" in MZ-DZ twin comparisons: an untenable premise of psychiatric genetics?
The comparison of MZ-DZ twins in behavioral genetics has produced what seems like irrefutable evidence for the heritability of many psychiatric disorders. But such research depends on the validity of the EEA--the "equal environments assumption"--as an underlying premise. In this paper, several empirical studies which support the EEA are critically reviewed in terms of methodology and the way data has been processed in a mathematical model called "path analysis". It turns out that studies investigating the EEA appear to be largely inadequate in terms of technique, as well as biased in the inferences drawn. Further, the "heritability" estimate--often taken to mean the influence of trait-specific genes--is merely a statistical abstraction derived from a matrix of correlations; this estimate encompasses many buried environmental effects so that "heritability" does not correspond to any underlying DNA structure. In conclusion, many MZ-DZ pedigree studies have dubious scientific value, given the non-viable premise of the EEA, as well as the misleading operational definition of what has been called "heritability".
Humans↗