PubMed Health⌕ Search

Biomedical subjects

M T Bradley

Publications and source records attributed to M T Bradley.

9 recordsLinked to original sources
PubMed2004

Diagnosing estimate distortion due to significance testing in literature on detection of deception.

Studies journals typically report or feature results significant by statistical test criterion. This is a bias that prevents obtaining precise estimates of the magnitude of any underlying effect. It is severe with small effect sizes and small numbers of measurements. To illustrate the problem and a diagnosis technique, results of published studies on the detection of deception are graphed. The literature contains large effect sizes affirming that deceptive responses in contrast to truthful responses are associated with more reactive Skin Resistance Responses. These effect sizes when graphed on the x-axis against n on the y-axis are distributed as funnel graphs. A subset of studies show support for predicted small to medium effects on different physiological measures, individual differences, and condition manipulations. These effect sizes graphed by sample ns follow negative correlations, suggesting that effect sizes from published values of t, F, and zeta are exaggerations.

Deception↗
PubMed2002

A Monte-Carlo estimation of effect size distortion due to significance testing.

A Monte-Carlo study was done with true effect sizes in deviation units ranging from 0 to 2 and a variety of sample sizes. The purpose was to assess the amount of bias created by considering only effect sizes that passed a statistical cut-off criterion of alpha = .05. The deviation values obtained at the .05 level jointly determined by the set effect sizes and sample sizes are presented. This table is useful when summarizing sets of studies to judge whether published results reflect an accurate appraisal of an underlying effect or a distorted estimate expected because significant studies are published and nonsignificant results are not. The table shows that the magnitudes of error are substantial with small sample sizes and inherently small effect sizes. Thus, reviews based on published literature could be misleading and especially so if true effect sizes were close to zero. A researcher should be particularly cautious of small sample sizes showing large effect sizes when larger samples indicate diminishing smaller effects.

Humans↗
PubMed1998

A control question test oriented towards students.

Control Question Tests were altered for 12 of 24 students who were examined with a polygraph about a mock crime which half of them had committed. The altered tests substituted control questions about students' cheating and plagiarism for the standard questions about crime issues. Responses to the altered tests were compared with those from tests using regular control questions which are usually about criminal issues. All tests were conducted by a professor. Detection scores derived from response magnitudes of skin resistance differed between innocent (M = 2.0) and guilty participants (M = -1.9). Guilt and innocence interacted with the type of test. Those examined with control questions oriented towards students scored as more innocent when actually innocent (M = 4.3) than guilty students examined with the student form (M = -3.0) or the crime form (M = -0.8) of the test and innocent students (M = -0.3) examined with control questions oriented towards crimes. The discussion is augmented by results from a direct analysis of magnitude of scores.

Crime↗
PubMed1996

The Control Question Test in polygraphic examinations with actual controls for truth.

Subjects who were guilty of a mock crime, innocent and informed of the details of the crime or innocent, and uninformed of the details were examined on the polygraph with a modified version of a Control Question Test, which normally contains questions which are incriminating, ambiguous, and likely to be answered with a lie. We challenged the necessity of using incriminating, ambiguous control questions which may demand a lie as an answer. Instead, we created and tested control questions that were unambiguous and answered truthfully. The results, with the modified questions, showed correct identification of 86% of the guilty subjects and 83 to 89% of the innocent subjects.

Adult↗
PubMed1993

Polygraph lie detection on real events in a laboratory setting.

This laboratory study dealt with real-life intense emotional events. Subjects generated embarrassing stories from their experience, then submitted to polygraph testing and, by lying, denied their stories and, by telling the truth, denied a randomly assigned story. Money was given as an incentive to be judged innocent on each story. An interrogator, blind to the stories, used Control Question Tests and found subjects more deceptive when lying than when truthful. Stories interacted with order such that lying on the second story was more easily detected than lying on the first. Embarrassing stories provide an alternative to the use of mock crimes to study lie detection in the laboratory.

Adult↗
PubMed1989

To what extent would women prefer chorionic villus sampling to amniocentesis for prenatal diagnosis?

Sixty-six pregnant women and 46 doctors were interviewed about their preferences for chorionic villus sampling (CVS) or amniocentesis for prenatal diagnosis in a hypothetical situation where the indication was late maternal age. The standard gamble method was used to calculate each individual's degree of preference for one procedure over the other (utility) expressed in terms of the risk of miscarriage associated with the preferred procedure that would be tolerated in order to have that procedure. Utilities for each group were calculated and compared. Pregnant women nominated a median utility for CVS of a miscarriage rate of 0.9%, while doctors nominated a median utility for CVS of a miscarriage rate of 1.2%. The difference between these utilities was not statistically significant. The method described in this study can enable potential consumers of a new procedure to provide the minimum 'clinically important difference' between a new procedure and an existing procedure necessary for calculation of the sample size in a controlled clinical trial.

Adolescent↗