Analysis Of Variance Design Research Articles

ANOVA MultiMedia: A Program for Teaching ANOVA Designs

A multimedia program for teaching analysis of variance (ANOVA) designs is described. The program tests students' understanding of independent groups and repeated measures variables, as well as the source and degrees of freedom columns of the ANOVA source table. The program uses colorful images, brief animations, interactive tasks, and immediate feedback. Information about availability is provided.

Transient evoked otoacoustic emissions: feasibility in the nursery.

To assess the feasibility of obtaining transient-evoked otoacoustic emissions (TEOAEs) directly in the nursery, to examine the nature of failures in this population, and to determine the time requirements for this test relative to the auditory brain stem response (ABR). TEOAEs and ABRs were recorded from 149 ears, at bedside, in the recovery or intensive care nursery. Subjects were selected according to the risk criteria developed by the Joint Committee on Infant Hearing (1991). Parametric and nonparametric statistical procedures were conducted to describe demographics and test results, to evaluate TEOAE measures by subgroups, and to compare infants by pass/fail outcome. Multivariate techniques were applied to the data to test implicit hypotheses regarding the source of TEOAE failures. In addition, the time required to carry out each test was evaluated in an analysis of variance design. 63.5% of the ears studied passed both procedures, 5% failed both, and 31.5% passed the ABR but failed the TEOAE. Although infants who failed the TEOAE had lower birthweights, shorter gestational periods, and were younger at the time of testing, no statistical differences were found for nursery (intensive care nursery/recovery), bed type (isolette/open crib), ear (left/right), or gender. Infants < 38 wk postconceptional age had smaller TEOAE responses and lower noise levels than did those > 37 wk. There were no differences in environmental noise levels (at the microphone) for passed or failed ears or whether infants were in isolettes or open cribs. Statistically prolonged ABR wave I latencies (Z-scores) at 60 and 30 dB among infants who passed the ABR but failed the TEOAE indicated some obstruction to the acoustic stimulus, whereas TEOAE noise measures neither distinguished these groups nor identified this condition. Three summary times involved in the performance of these tests were obtained: (a) actual test time, measured from start to completion of data collection with a stopwatch; (b) total test time, defined by the actual test time plus the respective preparation, setup, and cleanup (ABR) time; and (c) time reported by the computer system during intrinsic sampling. There were no significant differences in actual or total test times, but these values were distinguished from the machine time, which represented an underestimate of the time required to effectively conduct each test. These findings show that TEOAEs can be acquired in the nursery, but the high false-positive rate suggests that alternative or additional screening methods, for example, the ABR, must also be available. It must be recognized that TEOAE failures increase test time, thereby negating any savings relative to ABR screening alone.

Soft neurological signs in children: A normative study

A soft neurological signs (SNS) test battery (the Quick Neurological Screening Test; Mutti, Sterling, & Spalding, 1978) was individually administered to 233 normal children, ages 5 to 12 years. The participants’ characteristics were balanced according to (a) age (5–6 years, 7–8 years, 9–10 years, and 11–12 years); (b) socioeconomic status (SES; high and low); and (c) gender (boys and girls), yielding a 4 × 2 × 2 analysis of variance design with an average of 15 participants in each cell. In addition, an Academic Achievement Questionnaire was presented to the children's teachers. Normative data are presented. It was observed that (a) SNS decreased in frequency with age; (b) there were important individual differences in their frequency; (c) a significantly higher frequency of SNS was present in low‐SES than in high‐SES children, although these differences tended to decrease with age and schooling; (d) correlations between SNS and academic performance were in general low, but at least for some of these corr...

Controlled-Environment Evaluation of Fluorescent Strong Yellow-Green Pedestrian-Crossing Sign Prototypes

Fluorescent strong yellow-green (SYG) is one of the most unassigned colors in the Manual on Uniform Traffic Control Devices (MUTCD). FHWA has recently been considering the SYG color for nonmotorized application in the W11-2 series of signs in MUTCD. A number of field evaluation studies on the impact of these warning signs on conflicts and user preferences have been undertaken. A controlled-environment evaluation of the SYG pedestrian-crossing sign prototypes under five different natural light conditions is presented. Factors that contribute to different driver performance in response to microprismatic fluorescent SYG signs relative to the commonly used enclosed-bead yellow signs are identified. Target value and legibility distance of the prototypes in a simulated driving environment on a pool of test subjects are assessed. A factorial analysis of variance design was used to obtain unbiased estimates of these parameters. The general linear model was used iteratively to eliminate factors that did not affect the model significantly. It was found that SYG pedestrian-crossing prototypes had better performances in terms of their legibility distances than the enclosed-bead prototypes although the target value did not show any difference. The results indicate that use of fluorescent SYG signs is likely to increase driver awareness of warning signs.

Controlled-Environment Evaluation of Fluorescent Strong Yellow-Green Pedestrian-Crossing Sign Prototypes

Fluorescent strong yellow-green (SYG) is one of the most unassigned colors in the Manual on Uniform Traffic Control Devices (MUTCD). FHWA has recently been considering the SYG color for nonmotorized application in the W11-2 series of signs in MUTCD. A number of field evaluation studies on the impact of these warning signs on conflicts and user preferences have been undertaken. A controlled-environment evaluation of the SYG pedestrian-crossing sign prototypes under five different natural light conditions is presented. Factors that contribute to different driver performance in response to microprismatic fluorescent SYG signs relative to the commonly used enclosed-bead yellow signs are identified. Target value and legibility distance of the prototypes in a simulated driving environment on a pool of test subjects are assessed. A factorial analysis of variance design was used to obtain unbiased estimates of these parameters. The general linear model was used iteratively to eliminate factors that did not affect...

On emphasizing variance in the teaching of analysis of variance designs

Many common analysis of variance designs can be rapidly computed with the use of calculators and spreadsheet packages that have standard deviation functions. The calculator methods in this article free students from the tedium of calculating sums of squares and yet enable them to understand how sums of squares are obtained in ANOVA designs.

Effects of Induced Social Roles on the High School Personality Questionnaire

A one-way multivariate analysis of variance design with a control group (regular directions) and three treatment groups using induced social roles (Faking Good, Teacher, and Ideal Teacher) as independent variables and the High School Personality Questionnaire primary scores as dependent variables was used. Subjects were 384 male high school students from Grades 9 through 12. Within each classroom, students were randomly assigned to the four groups noted above. A broad pattern of differences in scores on primary and secondary personality dimensions were obtained. Significant differences between the control (standard directions) and the Faking Good, Teacher, and Ideal Teacher roles were obtained on three secondary and most of the primary personality dimensions. In several cases the ideal social role and neutral social role showed distinct differences from the more pervasive favorable impression role.

The Effects of Drug-Testing Program Characteristics on Applicants' Attitudes Toward Potential Employment

Abstract This study is an investigation of the effects of a mandatory versus probablecause drug-testing program and the level of restrictiveness of confidentiality policy on potential applicants' attitudes and intentions to apply for employment in an organization. One hundred fifty-six American students were randomly assigned to four experimental groups conforming to a 2 × 2 analysis of variance design: Type of Program (mandatory versus probable cause) x Level of Confidentiality (low versus high). A less restrictive confidentiality policy was associated with lower intentions to apply for employment relative to a more restrictive confidentiality policy. There were no perceived main or interactive effects involving a drug-testing program. Perceived procedural fairness partially mediated the effects of confidentiality on application intentions, and, in turn, an applicant's attitude toward the organization completely mediated the effects of perceived procedural fairness. These findings showed that procedural ...

Using a Group Editor with Alternative Communication Media in a Co-Authoring Environment

The writing efficiency of a type of computer-based groupware called a group editor was evaluated in six environments by factorially combining three communication media (audio only, audio plus video, and face-to-face) each with and without the group editor. Twelve subjects familiar with journalistic writing were pretested and matched into dyads. Subjects were trained with the group editor and wrote news articles based on a standard set of questions about actors and objects shown in short video clips. The quality of the writing task products was consistently high and showed no differences among conditions. Results of a balanced, within-subject analysis of variance design indicated that face-to-face conditions took significantly less time to complete than the other two communication conditions. When the group editor was used, however, all communication media were equal in terms of efficiency to face-to-face communications. Significantly less variance occurred in the audio plus video condition with the group editor than in the audio or the audio plus video conditions which did not have the group editor. Users preferred co-authoring with the group editor and considered writing trials with the group editor to be more productive. The results were discussed in terms of the benefits of using the group editor to increase the overall communication structure, to reduce variability of writing time among dyads, and to increase efficiency when face-to-face communication is not practical. Methodological procedures for matching subjects, using within-subject designs, and structuring a writing task were discussed as means of reducing team variability in writing efficiency when investigating computer-supported, cooperative work configurations.

Extreme Hypothermia Induced by a Synergism of Acute Limbic Seizures, Physical Restraint, and Acepromazine: Implications for Survival following Brain Injury

Core temperature was measured in rats 24 hr. after they had been assigned to one of 8 groups in a 3-way analysis of variance design that involved (1) induction of limbic seizures by a systemic injection of lithium/pilocarpine, (2) physical restraint, and (3) administration of acepromazine. An extraordinarily powerful interaction was noted among seizures, physical restraint, and acepromazine-produced hypothermia (24 degrees C) compared to the other 7 treatments (> 35 degrees C). The putative poikilothermic response is commensurate with the loss of mammalian behaviors that follow these seizures. Implications for survival during the acute stages of brain injury are suggested.

Experimental designs for aquatic mesocosm studies: A comparison of the “anova” and “regression” design for assessing the impact of tetrachlorophenol on zooplankton populations in limnocorrals

Two experimental designs were compared for assessing the impact of a commercial 2,3,4,6-tetrachlorophenol formulation (DIATOX®) on zooplankton abundance in a set of limnocorrals. The first design, based on blocked analysis of variance (ANOVA) procedures, used treatments of 0, 0.75, and 1.50 mg/L DIATOX with three replicates of each. The second design, based on regression procedures, had eight treatments ranging from 0 to 7.3 mg/L DIATOX with no replication. For the ANOVA experiment, a conventional ANOVA by day revealed significant impact on all zooplankton taxa investigated. Impact and time to recovery were greater with the 1.50-mg/L DIATOX treatment, and rotifers were more sensitive than macrozooplankton. A repeated-measures ANOVA, used to correct for pretreatment variability, revealed additional differences as significant. The ANOVA design permits easy assessment of impact at the chosen toxicant levels, but no conclusion can be drawn regarding the no-effect concentration (NEC) for major zooplankton taxa. For the regression experiment, linear dose—response relationships were used to estimate the NEC by inverse regression analysis. NECs for major zooplankton taxa and dominant species ranged from 0.28 mg/L DIATOX for immature Copepoda to 0.50 mg/L DIATOX for combined Cladocera. In 9 of 14 cases, however, the estimated EC50 ≤ NEC, an indication that an increased number of lower test concentrations were needed. The regression design offers a promising alternative to the common ANOVA design and allows for the estimation of specified end points such as the NEC and EC50.

Using Mystat to Analyze a Repeated-Measures Design

Laboratory experiments having a two-way analysis of variance design with repeated-measures on one factor are analyzed by MYSTAT. The key is in coding subjects as a factor and expressing the design as a three-way factorial with one subject per cell.

Cross-sectional and Longitudinal Analyses of Set in Relation to Age

Expectancy, age, and sex have been shown to affect reaction time (RT) in a variety of settings; however, the literature contains no studies exploring interactions among these factors using large-scale cross-sectional and longitudinal data. Thus, using data gathered over 17 years in the Baltimore Longitudinal Study of Aging (BLSA), the present study examine the effects of these factors on simple auditory reaction time for 1325 subjects (869 males, 456 females) on their first visit, 620 subjects (473 males, 147 females) for three visits in two-year intervals, and 307 subjects (277 males, 30 females) for five visits with two-year intervisit intervals. Across subjects and visits a fixed schedule of variable inter-stimulus intervals (ISIs) were employed ranging from 6 to 13 seconds (s), analyzed both as ISI and relative ISI (current ISI minus the previous ISI). Data were analyzed cross-sectionally according to decades in an Age8 × Sex2 × ISI8 mixed analysis of variance (ANOVA) design and an Age8 × Sex2 × Relative ISI6 mixed ANOVA design. Longitudinal analyses used similar designs but adjusted the number of age intervals examined as the sample decreased with visits. Results indicated that, as expected, with age came an overall slowing of RT, males were faster than females across decades. There was a relative ISI main effect demonstrating that across conditions, subjects reacted slower to cases in which long foreperiods were followed by short ones. The age by relative ISI interaction revealed that, like Botwinick, Brinley, and Birren (1957), the older subjects, compared to the young, were generally slower in responding and were caught “off guard” when long preparatory interval were followed by short ones. There was also a sex by relative ISI interaction such that the “being caught off-guard” was more pronounced in females than males. The ISI main effect took the form of a “U”, similar to what would be expected in the case of shorter ISIs, such that RT was slowest during the ISI conditions of 6 and 13 s and fastest in the middle. An age by sex by ISI interaction showed that older females had difficulty maintaining their state of readiness and the older males were especially slow when the stimulus occurred well before the general expected time. For longitudinal analyses, the pattern of visit 1–3 corresponds to the pattern of visit 1 almost exactly, but it also shows that there is slowing with repeat visit. In visit 1–5, there is a sex difference pattern of responding. Females have a pattern that progressively slows with repeat visit. But males tend to show increased RT between first visit and second visit, and then there is slight improvement with repeat visit. This study supports the notion that older individual have difficulty maintaining attention or preparedness and are particularly slow when arrival of the stimulus is unexpected or uncertained. A better understanding of set in relation to age has implication for task design, personnel selection, performance prediction, and accident prevention.

Contrast Analysis in Repeated Measures Analysis of Variance Designs: A SAS Implementation

The use of contrast analysis has been emphasized as an alternative to the two-stage sequence of omnibus F test followed by post-hoc pairwise comparisons. Instead, user contrasts tailored to the research can be formulated. Expressing each cell in a contrast in terms of the parameters of the means model is the key to determine the coefficients of the contrast. A three-way ANOVA with repeated measures on two factors is used to illustrate the technique. The code for the SAS Procedure GLM provides examples.

PAIRWISE COMPARISON PROCEDURES FOR ONE‐WAY ANALYSIS OF VARIANCE DESIGNS

ABSTRACTResearch in the behavioral and health sciences frequently involves the application of one‐factor analysis of variance models. The goal may be to compare several independent groups of subjects on a quantitative dependent variable or to compare measurements made on a single group of subjects on different occasions or under different conditions. In analyzing data of this kind, it is usually of interest to determine which pairs of population means are likely to differ. In this paper, the selection of pairwise multiple comparison procedures for one‐way analysis of variance designs is considered, following a discussion of Type I error and power issues as they apply to the testing of multiple hypotheses. Procedures are included which are appropriate when normality or variance homogeneity assumptions are violated. The focus is on procedures that are easy to understand and apply. Single‐step procedures are emphasized because of their simplicity and because they allow for the construction of confidence intervals.

A test of the “messages‐as‐fixed‐effect fallacy” argument: Empirical and theoretical implications of design choices

Jackson and Jacobs (1983) reopened a debate, already long resolved by statisticians and linguists, that dealt with treating language variables as random versus fixed effects in analysis of variance designs. Jackson and Jacobs extended the argument to suggest that designs were invalid unless they included multiple messages within single experiments. An ensuing debate has filled scarce journal space, often advancing specious arguments, but at times offering sensible alternatives to the position that there is an invariant messages‐as‐fixed‐effect fallacy in our literature. Despite the lengthy debate, there has been no empirical scrutiny of relevant claims. Presented here is the first empirical test of three alternative analytic models using data from a carefully controlled experiment.

Insulin and Dextrose Effects on Fixed-Interval Behavioral Thermoregulation in Albino Rats

This study is a systematic replication of the effects of insulin doses on operant behavior reinforced (in an earlier study) by fixed-ratio schedules of microwave (MW) reinforcement. In this study, insulin and dextrose doses were administered (ip) prior to fixed-interval 2-min. schedules of MW reinforcement in rats tested in a cold environment. Six Sprague-Dawley rats were conditioned to regulate their thermal environment with 5-sec. exposures of MW radiation (SAR = 0.34 Watts/kg/(mW/cm2) under the FI-2' schedules. Humulin-regular insulin and 50% solutions of dextrose were administered (ip) alternately with saline control sessions for 8-hr. durations. A within-subjects, repeated-measures 4 x 8 x 3 factorial analysis of variance design showed that insulin doses suppressed operant responding for heat, which confirmed the results of the earlier study under a different schedule. In addition, high doses of dextrose had similar suppressing effects on operant responding for heat. The data are interpreted in terms of the discriminative properties of increased thermogenesis produced by the insulin and dextrose doses. The suppressing effects were more pronounced for the first two hours, yet they persisted for approximately six hours of the 8-hr. sessions.

Overcoming Resistance to Persuasion VIA Contiguous Reinforcement and Repetition of Message

This study examined the effects of a contiguity learning model which posited one-trial learning based upon accurate knowledge of results as a reinforcer. A reconceptualization of reinforcing stimuli in the resistance to persuasion paradigm is experimentally tested. Repetition of the persuasive message was tested as part of the developed learning model and as a replication of earlier research. Knowledge of correct results and message repetition were manipulated to overcome resistance to persuasion created by pretreatment inoculation. Inoculation and persuasive attack messages were carefully created to control for a number of important message variables. A pilot study with 219 subjects tested effectiveness of the message. The actual experiment included 162 subjects in a three-way factorial analysis of variance design with inoculation, reinforcement, and repetition of message as predictor variables. Predicted main effects for each independent variable on posttest attitudes were found. Moreover, the pattern of posttest attitude means was exactly as posited. Discussion centered on the use of this learning model in research on persuasion.

Sexual Behavior, Church Attendance, and Permissive Beliefs among Unmarried Young Men and Women

Using a sample of 423 single men and women, we studied the frequency of sexual intercourse. It was predicted that to understand the conflicting findings about the effects of church attendance on young people's sexual behavior, other variables need to be considered. Accordingly, a five-way analysis of variance design, with independent variables of church attendance, sexual permissiveness, age, state, and gender, was used to study the dependent variables of sexual behavior. The only variable with a significant main effect was sexual permissiveness, not church attendance. However, there was a significant two-way interaction between church attendance and permissiveness and between permissiveness and gender. The means for permissive and nonpermissive females were similar, but for males the nonpermissive subjects had dramatically lower intercourse frequency. The interaction between church attendance and permissiveness resulted because nonpermissive males and females who attended church had the lowest frequency, but permissive subjects who attended church every week had one of the highest frequencies of sexual intercourse. The results are discussed in the context of other research on the relationship between church attendance and sexual behavior. We concluded that lower rates of sexual activity for this age group occur when there is a high level of church attendance but also a low level of sexual permissiveness.

Group versus individual instruction for teaching daily living skills to students with severe handicaps

The relative effectiveness of group and individual instruction formats for teaching daily living skills to students with severe handicaps was examined through an initial study (n = 23 students) and a replication (n = 47 students). The replication study also examined the acceptability of the group instruction procedures from the perspective of the participating teachers. Students were selected for participation who represented the range of students present in the classroom sites, including those students whom teachers identified as too difficult to teach in a group situation.Factorial repeated measures analysis of variance designs were used in both studies. Results indicated that there were no significant differences between the group and individual instruction formats for the four tasks targeted in the first study. These findings were replicated for the two tasks targeted in the second study. Virtually all of the teachers participating in the replication study reported the group instruction procedures to ...