CONFIDENCE CALIBRATION IN PERCEPTUAL AND INTELLIGENCE TASKS: ROLE OF NEED FOR CLOSURE, GENDER, AND ACADEMIC DOMAIN

Abstract

In the academic pursuit, students’ genuine confidence in themselves and their abilities is a notable achievement; however, misplaced confidence is risky and problematic. Therefore, educational psychologists pay special attention to understanding how well students calibrate while making judgments about their knowledge and performance in a domain, topic or task. Confidence calibration is the level of agreement between correctness and confidence in task performance. Seminal literature indicated that students are unequipped to evaluate their cognitive processing and unfit to analyze their knowledge gaps accurately and precisely. Consequently, they are prone to too much confidence or insufficient confidence while making judgments about their knowledge or task performance. This discrepancy between the student’s confidence and the accuracy of answers is called miscalibration or confidence bias, which is possible in two directions- over-confidence and under-confidence. Studies suggest that students are generally overconfident in their performance judgment; for example, 99% confident, while only 85% are correct. Previous theories delineated the patterns of individual differences displayed across tasks to explain the overconfidence effect. Hence, in this study, the individual differences-confidence-task performance paradigm was taken to explain the confidence bias. In this study NFC, gender and academic domain of the students were studied to understand their role in influencing undergraduate confidence calibration. This study was conducted in five phases. The consent for gathering information was performed in strict accordance with the institutional guidelines defined by the Human Ethics Committee of IIT Roorkee. After inspecting all the details of the data collection procedure, the ethical committee approved the research project. In the first phase of the research work, the psychometric properties of the NFC scale developed by Kruglanski et al. (2013) were checked on the Indian samples (N =450) by using CFA (parcels method) and found that the NFC scale is a reliable measure to be administered on the Indian undergraduate students. In the second phase, the undergraduate’s confidence calibration was analyzed on the perceptual task. The modified version of the sensory discrimination task initially developed by Baranski & Petrusic (1995) was used as a perceptual task. This task was an e-prime software designed, forced-choice two dimensional location comparison task that produced space errors of two difficulty levels by randomly changing certain stimulus configurations. Results suggested that students (N =189, Male = 105, Female = 84, Age = 18-25 years) deviated towards overconfidence in both perceptual task categories. In the third phase, the undergraduate’s confidence calibration was analyzed on the intelligence task. The modified version of the Raven Progressive Matrices test (Raven, 1936) was used as an intelligence task. The test items were categorized into two levels of difficulty. As Raven’s Test is strictly visual, the issues of potential language barriers and religious/cultural affiliations were circumvented. Results showed that students (N =102, Male = 55, Female = 47, Age = 18-25 years) deviated towards overconfidence in both intelligence task categories. In the fourth phase, the role of NFC, gender and academic domain differences was analyzed in the confidence calibration of the perceptual task. Screening of the data (N = 165) was done, and data were analyzed. Results indicated no significant role of NFC and academic domain but a partial role of gender in the perceptual task performance's accuracy, confidence and confidence bias. Similarly, in the fifth phase, the role of NFC, gender and academic domain differences was analyzed in the confidence calibration of the intelligence task. After data analysis (N =102), results were computed, which denied the significant interaction between the NFC and the accuracy, confidence and confidence bias. However, the partial interaction of gender and the academic domain was observed with the confidence ratings. Female undergraduates were found to be less confident than their male counterparts. Arts students were found to be comparatively more confident than science students. Interestingly, the confidence exceeded when the difficulty levels of the task increased. This study has implications in the classroom settings to make students more aware of their confidence in their knowledge.