The new and improved two-sample t test
Article Abstract:
This article considers the problem of comparing two independent groups in terms of some measure of location. It is well known that with Student's two-independent-sample t test, the actual level of significance can be well above or below the nominal level, confidence intervals can have inaccurate probability coverage, and power can be low relative to other methods. A solution to deal with heterogeneity is Welch's (1938) test. Welch's test deals with heteroscedasticity but can have poor power under arbitrarily small departures from normality. Yuen (1974) generalized Welch's test to trimmed means; her method provides improved control over the probability of a Type I error, but problems remain. Transformations for skewness improve matters, but the probability of a Type I error remains unsatisfactory in some situations. We find that a transformation for skewness combined with a bootstrap method improves Type I error control and probability coverage even if sample sizes are small.
Publication Name: Psychological Science
Subject: Psychology and mental health
ISSN: 0956-7976
Year: 2004
User Contributions:
Comment about this article or add new information about this topic:
Learning to time (LET) or scalar expectancy theory (SETt)? A critical test of two models of timing
Article Abstract:
A study was conducted to contrast the predictions of two major frameworks of timing. A variation of temporal bisection procedure was utilized to carry out the study. Novel combinations of choice keys were used, and preference for a specific key was evaluated as a function of signal duration. Results indicated that behaviors observed during temporal cues are not unrelated to, or a mere expression of, an internal clock.
Publication Name: Psychological Science
Subject: Psychology and mental health
ISSN: 0956-7976
Year: 1999
User Contributions:
Comment about this article or add new information about this topic:
Rethinking how family researchers model infrequent outcomes: a tutorial on count regression and zero-inflated models
Article Abstract:
Statistical methods like Poisson regression, negative binomial, zero-inflated Poisson and zero-inflated negative binomial regression models are discussed. These methods are specifically developed for skewed event data and can be used by family researchers who use the ordinary least-squares regressions.
Publication Name: Journal of Family Psychology
Subject: Psychology and mental health
ISSN: 0893-3200
Year: 2007
User Contributions:
Comment about this article or add new information about this topic:
- Abstracts: Gender and depressive symptoms: parents' sharing of instrumental and expressive tasks when their children are young
- Abstracts: A Note on Tipping and Employee Peceptions and Attitudes. Subjective metaphysics and learning from experience: the causal psychology of rational choice
- Abstracts: The role of past states in determining reference points for policy decisions. Person-organization fit, job choice decisions, and organizational entry
- Abstracts: American and German students' knowledge, perceptions, and behaviors with respect to over-the-counter pain relievers