China
Africa
Explore chapters and articles related to this topic
The neuropsychological approach
Published in Stanley Berent, James W. Albers, Neurobehavioral Toxicology, 2012
Stanley Berent, James W. Albers
A test's reliability is determined through scientific enquiry and statistical analysis. The main statistical tool used to establish a test's reliability is correlation, and the resultant statistic can be generally referred to as the reliability coefficient. Reliability is of several types, each reflecting a different aspect of the test's internal consistency of measurement. A few of the major types of reliability are test–retest, alternate form, and split-half. Tests also vary with regard to the form of stimulus presentation. Various specific correlation methods have been developed to be used in establishing a test's reliability, depending on the type of reliability or test format. Cronbach's coefficient alpha (Cronbach, 1951) was designed for use with multiple-choice-formatted tests, for instance. Dichotomous items (e.g., true or false items) will call for a statistical test such as the Kuder–Richardson formula 20 (Kuder & Richardson, 1937), while split-half formatted tests might use the Spearman–Brown coefficient of correlation (Walker & Lev, 1953). Some types of tests do not lend well to establishing internal consistency. This is true, for example, of most tasks that require speed over accuracy for successful completion (such as the finger-tapping task used in the Halstead–Reitan Battery) (Reitan & Wolfson, 1993; Berent & Trask, 2000).
Validity and reliability
Published in Louis Cohen, Lawrence Manion, Keith Morrison, Research Methods in Education, 2017
Louis Cohen, Lawrence Manion, Keith Morrison
The issue here is that results as well as instruments should be reliable. Reliability is also addressed by: calculating coefficients of reliability, split-half techniques, the Kuder-Richardson formula, parallel/equivalent forms of a test, test/re-test methods, the alpha coefficient;calculating and controlling the standard error of measurement;increasing the sample size (to maximize the range and spread of scores in a norm-referenced test), though criterion-referenced tests recognize that scores may bunch around the high level (in mastery learning, for example), i.e. the range of scores might be limited, thereby lowering the correlation coefficients that can be calculated;increasing the number of observations made and items included in the test (in order to increase the range of scores);ensuring effective domain sampling of items in tests based on item response theory (a particular issue in Computer adaptive testing, introduced below (Thissen, 1990));ensuring effective levels of item discriminability and item difficulty. Reliability not only has to be achieved but has to be seen to be achieved, particularly in ‘high-stakes’ testing (where a lot hangs on the results of the test, e.g. entrance to higher education or employment). Hence the procedures for ensuring reliability must be transparent. The difficulty here is that the more one moves towards reliability as defined above, the more the test will become objective, the more students will be measured as though they are standardized objects, and the more the test will become decontextualized.
Reliability I: Classical methods
Published in Claudio Violato, Assessing Competence in Medicine and Other Health Professions, 2018
Advanced Organizers Reliability is a necessary but not sufficient condition for validity. Reliability is measured by an index called the reliability coefficient, rxx.Classical methods of reliability are based on classical test theory. A test score is composed of the “real” score or true score plus error of measurement: X (Observed score) = T (True score) + e (error of measurement).There are four basic methods for deriving rxx: test–retest, parallel forms, split-half, and internal consistency. Each method focuses on a somewhat different aspect of reliability.The Spearman–Brown prophecy formula describes the relationship between test length and reliability. Thus, once you know the reliability of a particular test, you can “prophesize” the effect on the reliability by either increasing the test length or decreasing it.The most practical and simplest method of determining reliability for classroom use is the Kuder–Richardson formula 21 (KR21), which is a general internal consistency method.The standard error of measurement (Se) translates the test’s reliability onto the actual test scale to estimate the error involved in actual scores that students receive. The Se is most useful for interpreting an individual’s performance on the test.The score a student actually obtains on a test is composed of the true score plus error of measurement. True scores are hypothetical entities estimated by the use of the Se and a particular confidence interval. Thus, the true score falls within a band of error as circumscribed by the Se with a specific degree of confidence.By convention, three confidence intervals are commonly used: 68% (±1 Se), 95% (±2 Se), and 99% (±3Se).There are several common factors that directly influence a test’s reliability. These include the test’s length, the heterogeneity of the test-takers, whether or not the test is speeded, the length of the time interval in test–retest and parallel forms at different times, the difficulty of the test, and the objectivity of the scoring of the test.
The Importance of Rest in Trauma Services: Perhaps We Should Consider Naps!
Published in Alcoholism Treatment Quarterly, 2023
Dana C Branson, Elizabeth D. Arlington, Christopher S. Bradley
A quantitative survey consisting of the Adverse Childhood Events (ACE) questionnaire (Felitti et al., 1998; Reavis et al., 2013), 10 additional yes-or-no questions that evaluated potential adult trauma, a single question that assessed SUD status, and five demographic items was completed by a total of 47 participants (n = 47). To maintain the anonymity of participants, no personally identifying information was collected as part of the quantitative survey. The ACE questionnaire was chosen in part because it has high reliability rates as shown through internal consistency scores typically above .80 (Bethell et al., 2017; Larkin, 2019; Meinck et al., 2017). ACE scores were computed by summing the 10 ACE scale items. It should be noted that each ACE item used a dichotomous yes or no answer format (Felitti et al., 1998; Reavis et al., 2013), which means that ACE scores ranged from a low of 0 to a high of 10. The 10 yes-or-no questions that evaluated potential adult trauma that were developed for use in the current investigation were also summated into a single scale. This scale, which will be referred to as the Adult Trauma Events (ATE) scale, ranged from a low of 0 to a high of 10.1 Because both the ACE and ATE scales use a yes or no answer format for all questions, computation of the Kuder-Richardson Formula 20 (KR-20) is recommended to demonstrate the α reliability of each scale (Cortina, 1993). The KR-20 α score for the ACE is .84, while the KR-20 α score for the ATE is .78.
“Why People Gotta be so Judgy?”: The Importance of Agency-Wide, Non-judgmental Approach to Client Care
Published in Alcoholism Treatment Quarterly, 2022
Dana C. Branson, Jocelyn S. Martin, Olivia E. Westbrook, River J. Ketcherside, Christopher S. Bradley
Table 1 presents descriptive statistics for the summated ACE scale and the 15 additional yes or no questions that assessed potential trauma experienced as an adult. The death of a close peer or loved one was the number one trauma experienced by respondents (91.5%), followed by emotional abuse (80.9%), divorce (74.5%) and physical abuse (66.0%). Nine out of ten respondents (91.5%) reported that they are struggling with a medical issue, and three out of four respondents (74.5%) noted that they are struggling with a mental health issue. The average amount of adverse childhood events for the sample, which ranged from zero events to 10 events, was about five and a half events (M = 5.60), with a standard deviation of roughly three events (SD = 3.06). Given that the ACE scale uses dichotomous answer formats for each of the ten questions within the summated scale (Felitti et al., 1988; Reavis et al., 2019), computation of the Kuder-Richardson Formula 20 (KR-20) is preferred over Cronbach alpha to demonstrate reliability of the scale (Cortina, 1993). The KR-20 for the ACE was .84.
Factors influencing preventive behaviors for Zika virus infection in female nursing students: A cross-sectional study
Published in Contemporary Nurse, 2020
The questionnaire used to evaluate the level of knowledge of the ZIKV was developed based on guidelines provided by Painter et al. (2017), the CDC (2016), and the Kuder and Richardson (1937). The questionnaire contained 17 items, including data on the causative agent and mechanisms of viral transmission (three items), symptoms (four items), complications (five items), epidemiology (two items), and treatment and prevention methods (three items). Each correct answer corresponded to one point, each wrong answer corresponded to zero points, and the total achievable score was 100 points: the higher the score, the higher the level of knowledge. The final CVI was 0.95. The reliability score of the questionnaire in the pilot study was Kuder–Richardson Formula 20 (KR 20; Kuder & Richardson, 1937) = .66 and that of the main study was KR 20 = .72.