Utilisation des intervalles de confiance au lieu des tests de signification : aspects épistémologiques et pratiques

The use of confidence intervals instead of significance tests is strongly recommended by the fifth edition of the manual of the American Psychological Association (2001). This possibility as well as other improvements in statistical practice are discussed in the framework of the major theoretical options subtending statistical inference and the way they have been applied in psychology for about 50 years. First, the suggestion of a complete ban on statistical testing is examined and rejected. Next, a procedure consisting in measuring the fit of two competing models based on the likelihood ratio is judged interesting and commendable. Finally, the superiority of an approach based on confidence intervals instead of significance tests is assessed and illustrated by its application to an experimental study aiming to demonstrate the absence instead of the presence of an effect of the independent variable.     

老年人错误记忆的情绪效价和时间效应

基于信号检测论的框架, 考察不同的情绪效价以及时间间隔如何交互影响老年人在图片再认任务中的辨别力(d′)和判断标准(β), 进而影响其错误记忆。以21名老年人(女性14名)为被试, 平均年龄67.17 ± 5.03岁。根据中国老年人对国际情绪图片库(International Affective Picture System, IAPS)中图片评定的情绪参数, 选出积极、消极和中性图片各60张作为学习材料。另选积极、消极和中性图片各30张作为每次再认测验的干扰材料, 并且两次再认测验的干扰材料不同。短时间隔条件让老年人在图片记忆编码半个小时后完成对图片的再认测验; 长时间隔条件则为在三周后完成对图片的再认测验。结果发现: 1)在短时间隔条件下, β和d′共同影响虚报率; 而在长时间隔条件下, 只有β影响虚报率, d′不影响虚报率; 2)无论是在短时还是长时间隔条件下, 积极图片与消极图片的判别力d′都没有显著差别; 3)在短时间隔条件下, 老年人对消极图片的β更低、虚报率更高; 在长时间隔条件下, 老年人对积极图片的β更低、虚报率更高。研究结果表明: 老年人的错误记忆受判别力d′和判断标准β的影响。情绪效价通过作用于老年人的反应倾向(判断标准β)、而非记忆质量(判别力d′)来影响其错误记忆。时间间隔可以调节情绪效价对老年人判断标准β和错误记忆的作用, 使之随时间发生反转。老年人的“积极效应”在错误记忆中可能表现为随着时间的流逝, 老年人更愿意将积极信息报告为经历或学习过的信息。     

The role of temporal intervals on reinforcer accumulation

Animals accumulate reinforcers when they forgo the opportunity to consume available food in favor of acquiring additional food for later consumption. Laboratory research has shown that reinforcer accumulation is facilitated when an interval (either spatial or temporal) separates earning from consuming reinforcers. However, there has been no systematic investigation on the interval separating consuming reinforcers from earning additional reinforcers. This oversight is problematic because this second interval is an integral part of much of the previous research on reinforcer accumulation. The purpose of the current study was to determine the independent contributions of these two temporal intervals on reinforcer accumulation in rats. Each left lever press earned a single food pellet; delivery of the accumulated pellet(s) occurred upon a right lever press. Conditions varied based on the presence of either an intertrial interval (ITI) that separated pellet delivery from the further opportunity to accumulate more pellets, or a delay‐to‐reinforcement that separated the right lever press from the delivery of the accumulated pellet(s). Delay and ITI values of 0, 5, 10 and 20 s were investigated. The delay‐to‐reinforcement conditions produced greater accumulation relative to the ITI conditions, despite accumulation increasing the density of reinforcement more substantially in the ITI conditions. This finding suggests that the temporal separation between reinforcer accumulation and subsequent delivery and consumption was a more critical variable in controlling reinforcer accumulation.     

Explaining the persistence of false memories: a proposal based on associative activation and thematic extraction

This study aimed to analyse the effect of retention intervals on associative and thematic false memories. Two experiments, using two types of critical items that were either associatively or thematically related to studied material, were conducted. In both experiments, one group of participants performed a recognition test immediately after the presentation of lists, and another group performed the task one week later. In Experiment 1, the recognition test consisted of pairs of items with four response alternatives (both items had been presented, only the left item had been presented, only the right item had been presented or none of the items had been presented). Critical items were also manipulated so that they were either presented in or absent from the list. In Experiment 2, a standard recognition test that differed in the mode of presentation was used: self-paced or speeded response. Both experiments showed that associative critical items were more recognised than thematic critical items in the immediate condition. However, whereas associative critical items decayed after a one-week delay, thematic critical items were similarly recognised at both retention intervals. The findings of the present study suggest that each type of process – associative and thematic – behave differently over time.     

与零假设检验有关的统计分析流程

人们熟知的零假设显著性检验,受到一次次质疑与辩护,地位并未动摇,报告检验结果仍然是统计分析的习惯做法。不过,其局限性促使研究者探寻更多的统计方法如区间估计、效应量分析、检验力分析等。本文先介绍假设检验与置信区间的关系;然后讨论检验力与两类错误率和效应量的关系;最后在理顺上述统计方法的基础上,提供一个可操作的统计分析流程。     

Instructions as discriminative stimuli (2): A within‐subject examination of the effect of differential reinforcement on establishing novel instructional control1

Abstract: Six undergraduates were exposed to a fixed‐ratio schedule with an instruction to respond slowly and to a differential‐reinforcement‐of‐low‐rate schedule with an instruction to respond rapidly when a white circle was presented on a display monitor. When a yellow circle was presented, however, the subjects were exposed to the fixed‐ratio schedule with the instruction to respond rapidly and to the differential‐reinforcement‐of‐low‐rate schedule with the instruction to respond slowly. Following this, a fixed‐interval schedule was in effect during those stimuli and instructions. Under the white circle, response rates were higher with the instruction to respond slowly than with the instruction to respond rapidly during the fixed‐interval schedule. Such control by instructions was not observed under the yellow circle. A previous study examined establishment of novel instructional control by between‐subject comparisons and found that for three of four subjects ( Okouchi, 1999 ). In contrast, the present results demonstrate the instructional control through within‐subject comparisons for all six subjects.     

Studies of wheel-running reinforcement: parameters of Herrnstein's (1970) response-strength equation vary with schedule order

Six male Wistar rats were exposed to different orders of reinforcement schedules to investigate if estimates from Herrnstein's (1970) single-operant matching law equation would vary systematically with schedule order. Reinforcement schedules were arranged in orders of increasing and decreasing reinforcement rate. Subsequently, all rats were exposed to a single reinforcement schedule within a session to determine within-session changes in responding. For each condition, the operant was lever pressing and the reinforcing consequence was the opportunity to run for 15 s. Estimates of k and R(O) were higher when reinforcement schedules were arranged in order of increasing reinforcement rate. Within a session on a single reinforcement schedule, response rates increased between the beginning and the end of a session. A positive correlation between the difference in parameters between schedule orders and the difference in response rates within a session suggests that the within-session change in response rates may be related to the difference in the asymptotes. These results call into question the validity of parameter estimates from Herrnstein's (1970) equation when reinforcer efficacy changes within a session.     

Drug discrimination under two concurrent fixed-interval fixed-interval schedules

Pigeons were trained to discriminate 5.0 mg/kg pentobarbital from saline under a two-key concurrent fixed-interval (FI) 100-s FI 200-s schedule of food presentation, and later tinder a concurrent FI 40-s FI 80-s schedule, in which the FI component with the shorter time requirement reinforced responding on one key after drug administration (pentobarbital-biased key) and on the other key after saline administration (saline-biased key). After responding stabilized under the concurrent FI 100-s FI 200-s schedule, pigeons earned an average of 66% (after pentobarbital) to 68% (after saline) of their reinforcers for responding under the FI 100-s component of the concurrent schedule. These birds made an average of 70% of their responses on both the pentobarbital-biased key after the training dose of pentobarbital and the saline-biased key after saline. After responding stabilized under the concurrent FI 40-s FI 80-s schedule, pigeons earned an average of 67% of their reinforcers for responding under the FI 40 component after both saline and the training dose of pentobarbital. These birds made an average of 75% of their responses on the pentobarbital-biased key after the training dose of pentobarbital, but only 55% of their responses on the saline-biased key after saline. In test sessions preceded by doses of pentobarbital, chlordiazepoxide, ethanol, phencyclidine, or methamphetamine, the dose-response curves were similar under these two concurrent schedules. Pentobarbital, chlordiazepoxide, and ethanol produced dose-dependent increases in responding on the pentobarbital-biased key as the doses increased. For some birds, at the highest doses of these drugs, the dose-response curve turned over. Increasing doses of phencyclidine produced increased responding on the pentobarbital-biased key in some, but not all, birds. After methamphetamine, responding was largely confined to the saline-biased key. These data show that pigeons can perform drug discriminations under concurrent schedules in which the reinforcement frequency under the schedule components differs only by a factor of two, and that when other drugs are substituted for the training drugs they produce dose-response curves similar to the curves produced by these drugs under other concurrent interval schedules.     

Pharmacological and physiological aspects of sexual exhaustion in male rats

The present article reviews the current findings on the interesting phenomenon of sexual satiety. Knut Larsson in 1956 reported on the development of sexual exhaustion in the male rat after repeated copulation. We have studied the process and found the following results. (1) One day after 4 hours of ad libitum copulation, two-thirds of the population showed complete inhibition of sexual behavior, while the other third displayed a single ejaculatory series from which they did not recover. (2) Several pharmacological treatments, including 8-OH-DPAT, yohimbine, naloxone and naltrexone, reverse this sexual satiety, indicating that the noradrenergic, serotonergic and opiate systems are involved in this process. Indeed, direct neurochemical determinations showed changes in various neurotransmitters during sexual exhaustion. (3) Given enough stimulation, by changing the stimulus female, sexual satiety was prevented, suggesting that there are motivational components of the sexual inhibition that characterizes sexual exhaustion. (4) The GABA antagonist bicuculline, or the electrical stimulation of the medial preoptic area, did not reverse sexual exhaustion. These data suggest, on the one hand, that sexual exhaustion and the postejaculatory interval (which is shortened by bicuculline administration) are not mediated by similar mechanisms and, on the other, that the medial preoptic area does not regulate sexual satiety. (5) The androgen receptor density in brain areas closely related to the expression of masculine sexual behavior, such as the medial preoptic nucleus, was drastically reduced in sexually exhausted animals. Such reduction was specific to certain brain areas and was not related to changes in the levels of androgens. These results suggest that changes in brain androgen receptors account for the inhibition of sexual behavior present during sexual exhaustion. (6) The recovery process of sexual satiety after 4 hours of ad libitum copulation reveals that, after 4 days, only 63% of the males are able to show sexual behavior while after 7 days all animals display copulatory activity.     

Exact distributions of intraclass correlation and Cronbach's alpha with Gaussian data and general covariance

Intraclass correlation and Cronbach's alpha are widely used to describe reliability of tests and measurements. Even with Gaussian data, exact distributions are known only for compound symmetric covariance (equal variances and equal correlations). Recently, large sample Gaussian approximations were derived for the distribution functions. New exact results allow calculating the exact distribution function and other properties of intraclass correlation and Cronbach's alpha, for Gaussian data with any covariance pattern, not just compound symmetry. Probabilities are computed in terms of the distribution function of a weighted sum of independent chi-square random variables. NewF approximations for the distribution functions of intraclass correlation and Cronbach's alpha are much simpler and faster to compute than the exact forms. Assuming the covariance matrix is known, the approximations typically provide sufficient accuracy, even with as few as ten observations. Either the exact or approximate distributions may be used to create confidence intervals around an estimate of reliability. Monte Carlo simulations led to a number of conclusions. Correctly assuming that the covariance matrix is compound symmetric leads to accurate confidence intervals, as was expected from previously known results. However, assuming and estimating a general covariance matrix produces somewhat optimistically narrow confidence intervals with 10 observations. Increasing sample size to 100 gives essentially unbiased coverage. Incorrectly assuming compound symmetry leads to pessimistically large confidence intervals, with pessimism increasing with sample size. In contrast, incorrectly assuming general covariance introduces only a modest optimistic bias in small samples. Hence the new methods seem preferable for creating confidence intervals, except when compound symmetry definitely holds. An earlier version of this paper was submitted in partial fulfillment of the requirements for the M.S. in Biostatistics, and also summarized in a presentation at the meetings of the Eastern North American Region of the International Biometric Society in March, 2001. Kistner's work was supported in part by NIEHS training grant ES07018-24 and NCI program project grant P01 CA47 982-04. She gratefully acknowledges the inspiration of A. Calandra's “Scoring formulas and probability considerations” (Psychometrika, 6, 1–9). Muller's work supported in part by NCI program project grant P01 CA47 982-04.