Meintest du stage?
Autorenhinweise des Verlags Taylor & Francis sind hier abrufbar Sage Journals Das
Effective science communication (SciComm) is crucial, but training scalability remains challenging. We explored whether generative AI (GenAI) could provide feedback to enhance SciComm strategies. In an online iterative distillation exercise, SciComm trainees (N = 78) condensed their research. An experimental group (n = 41) received jargon-oriented and GenAI feedback; controls (n = 37) received only jargon feedback. Participants preferred revised texts, with slightly higher preference in the GenAI group. SciComm-based rubric assessment revealed GenAI-supported texts significantly improved in SciComm strategies, particularly connecting science to everyday life and narrative use. Findings highlight GenAI’s potential to enhance SciComm content and scalable feedback, supporting its careful integration into training.
Englisch Erschienen in Science Communication Seiten 27 Herausgeber (Verlag) SAGE
Effective science communication (SciComm) is crucial, but training scalability remains challenging. We explored whether generative AI (GenAI) could provide feedback to enhance SciComm strategies. In an online iterative distillation exercise, SciComm trainees (N = 78) condensed their research. An experimental group (n = 41) received jargon-oriented and GenAI feedback; controls (n = 37) received only jargon feedback. Participants preferred revised texts, with slightly higher preference in the GenAI group. SciComm-based rubric assessment revealed GenAI-supported texts significantly improved in SciComm strategies, particularly connecting science to everyday life and narrative use. Findings highlight GenAI’s potential to enhance SciComm content and scalable feedback, supporting its careful integration into training.
Englisch Erschienen in Science Communication Seiten 27 Herausgeber (Verlag) SAGE
Holland’s RIASEC theory posits that career choices are guided by the desire to establish congruence between individual interests and environmental activity opportunities. Furthermore, the theory states that most individuals exhibit interests that are consistent with the circular ordering of RIASEC domains, thereby rendering specific interest constellations unlikely. Therefore, as activity opportunities provided by many environments are inconsistent with the circular RIASEC order, the question emerges as to which individuals they attract. In this article, we examine the plausibility of three different scenarios that assert inconsistent environments (A) attract individuals with similarly inconsistent interests; (B) attract individuals with consistent, but only partially congruent interests; and (C) attract individuals with overall strong interests. Data collected within three university majors (N = 553)—two consistent (emphasizing either R or S) and one inconsistent (emphasizing both R and S)—support Scenario B, suggesting that inconsistent environments are linked to lower congruence and greater interest diversity.
Erschienen in Journal of Career Development, 52(4) Seiten 427–448 Herausgeber (Verlag) SAGE
Holland’s RIASEC theory posits that career choices are guided by the desire to establish congruence between individual interests and environmental activity opportunities. Furthermore, the theory states that most individuals exhibit interests that are consistent with the circular ordering of RIASEC domains, thereby rendering specific interest constellations unlikely. Therefore, as activity opportunities provided by many environments are inconsistent with the circular RIASEC order, the question emerges as to which individuals they attract. In this article, we examine the plausibility of three different scenarios that assert inconsistent environments (A) attract individuals with similarly inconsistent interests; (B) attract individuals with consistent, but only partially congruent interests; and (C) attract individuals with overall strong interests. Data collected within three university majors (N = 553)—two consistent (emphasizing either R or S) and one inconsistent (emphasizing both R and S)—support Scenario B, suggesting that inconsistent environments are linked to lower congruence and greater interest diversity.
Erschienen in Journal of Career Development, 52(4) Seiten 427–448 Herausgeber (Verlag) SAGE
While there is ample theoretical and empirical evidence detailing which conditions benefit learning from one’s own errors, the evidence on learning from others’ errors has not yet been synthesized. In this meta-analysis, we examine the overall impact of erroneous examples on learning and the effects of potential moderating variables based on a novel framework. Following the robust variance estimation method, we synthesized findings from 42 papers (177 effect sizes) comparing erroneous examples with correct examples or problem-solving in experimental studies. The results revealed a statistically significant but weak effect of erroneous examples on learning (g = .136). Further analysis indicated a statistically significant moderating effect of the design of error-explanation activities. Specifically, providing self-explanation prompts or instructional explanations enhanced learning from erroneous examples more than not providing any error explanations. Our findings draw attention to the design of error explanation activities as well as several areas for future research.
Originalsprache Englisch Erschienen in Review of Educational Research Herausgeber (Verlag) Sage
While there is ample theoretical and empirical evidence detailing which conditions benefit learning from one’s own errors, the evidence on learning from others’ errors has not yet been synthesized. In this meta-analysis, we examine the overall impact of erroneous examples on learning and the effects of potential moderating variables based on a novel framework. Following the robust variance estimation method, we synthesized findings from 42 papers (177 effect sizes) comparing erroneous examples with correct examples or problem-solving in experimental studies. The results revealed a statistically significant but weak effect of erroneous examples on learning (g = .136). Further analysis indicated a statistically significant moderating effect of the design of error-explanation activities. Specifically, providing self-explanation prompts or instructional explanations enhanced learning from erroneous examples more than not providing any error explanations. Our findings draw attention to the design of error explanation activities as well as several areas for future research.
Originalsprache Englisch Erschienen in Review of Educational Research Herausgeber (Verlag) Sage
In longitudinal assessments, tests are frequently used to estimate trends over time. However, when item parameters lack invariance, time-point comparisons can be distorted, necessitating appropriate statistical methods to achieve accurate estimation. This study compares trend estimates using the two-parameter logistic (2PL) model under item parameter drift (IPD) across five trend-estimation approaches for two time points: First, concurrent calibration, which jointly estimates item parameters across multiple time points. Second, fixed calibration, which estimates item parameters at a single time point and fixes them at the other time point. Third, robust linking with Haberman and Haebara as linking methods withLporL0losses. Fourth, non-invariant items are detected using likelihood-ratio tests or the root mean square deviation statistic with fixed or data-driven cutoffs, and trend estimates are then recomputed using only the detected invariant items under partial invariance. Fifth, regularized estimation under a smooth Bayesian information criterion (SBIC) is applied, shrinking small or null IPD effects toward zero while estimating all others as nonzero. Bias and relative root mean square error (RMSE) were evaluated for the mean and SD at T2. An empirical example using synthetic longitudinal reading data, applying the trend-estimation approaches, is provided. The results indicate that the regularized estimation with SBIC performed best across conditions, maintaining low bias and RMSE, followed by robust linking methods. Specifically, Haberman linking with theL0loss function showed superior performance under unbalanced IPD, outperforming the partial invariance approaches. Concurrent and fixed calibration showed the poorest trend recovery under unbalanced IPD conditions.
Erschienen in Educational and Psychological Measurement Herausgeber (Verlag) SAGE
In longitudinal assessments, tests are frequently used to estimate trends over time. However, when item parameters lack invariance, time-point comparisons can be distorted, necessitating appropriate statistical methods to achieve accurate estimation. This study compares trend estimates using the two-parameter logistic (2PL) model under item parameter drift (IPD) across five trend-estimation approaches for two time points: First, concurrent calibration, which jointly estimates item parameters across multiple time points. Second, fixed calibration, which estimates item parameters at a single time point and fixes them at the other time point. Third, robust linking with Haberman and Haebara as linking methods withLporL0losses. Fourth, non-invariant items are detected using likelihood-ratio tests or the root mean square deviation statistic with fixed or data-driven cutoffs, and trend estimates are then recomputed using only the detected invariant items under partial invariance. Fifth, regularized estimation under a smooth Bayesian information criterion (SBIC) is applied, shrinking small or null IPD effects toward zero while estimating all others as nonzero. Bias and relative root mean square error (RMSE) were evaluated for the mean and SD at T2. An empirical example using synthetic longitudinal reading data, applying the trend-estimation approaches, is provided. The results indicate that the regularized estimation with SBIC performed best across conditions, maintaining low bias and RMSE, followed by robust linking methods. Specifically, Haberman linking with theL0loss function showed superior performance under unbalanced IPD, outperforming the partial invariance approaches. Concurrent and fixed calibration showed the poorest trend recovery under unbalanced IPD conditions.
Erschienen in Educational and Psychological Measurement Herausgeber (Verlag) SAGE
Wissenschaftlerinnen und Wissenschaftler haben ein Papier geschrieben. In dem Papier geht es um KI-Sprachsysteme und wie man mit diesen gut Lernen kann.
Die Experten sagen: Wir sollen die Computer-Programme nicht verbieten.
