Subjects: Other Disciplines >> Synthetic discipline submitted time 2023-10-09 Cooperative journals: 《心理学报》
Abstract： Latent growth models (LGMs) are a powerful tool for analyzing longitudinal data, and have attracted the attention of scholars in psychology and other social science disciplines. For a latent variable measured by multiple indicators, we can establish both a univariate LGM (also called first-order LGM) based on composite scores and a latent variable LGM (also called second-order LGM) based on indicators. The two model types are special cases of the first-order and second-order factor models respectively. In either case, we need to scale the factors, that is, to specify their origin and unit. Under the condition of strong measurement invariance across time, the estimation of growth parameters in second-order LGMs depends on the scaling method of factors/latent variables. There are three scaling methods: the scaled-indicator method (also called the marker-variable identification method), the effect-coding method (also called the effect-coding identification method), and the latent-standardization method. The existing latent-standardization method depends on the reliability of the scaled-indicator or the composite scores at the first time point. In this paper, we propose an operable latent-standardization method with two steps. In the first step, a CFA with strong measurement invariance is conducted by fixing the mean and variance of the latent variable at the first time point to 0 and 1 respectively. In the second step, estimated loadings in the first step are employed to establish the second-order LGM. If the standardization is based on the scaled-indicator method, the loading of the scaled-indicator is fixed to that obtained in the first step, and the intercept of the scaled-indicator is fixed to the sample mean of the scaled-indicator at the first time point. If the standardization is based on the effect-coding method, the sum of loadings is constrained to the sum of loadings obtained in the first step, and the sum of intercepts is constrained to the sum of the sample mean of all indicators at the first time point. We also propose a first-order LGM standardization procedure based on the composite scores. First, we standardize the composite scores at the first time point, and make the same linear transformation of the composite scores at the other time points. Then we establish the first-order LGM, which is comparable with the second-order LGM scaled by the latent-standardization method. The scaling methods of second-order LGMs and their comparable first-order LGMs are systematically summarized. The comparability is illustrated by modeling the empirical data of a Moral Evasion Questionnaire. For the scaled-indicator method, second-order LGMs and their comparable first-order LGMs are rather different in parameter estimates (especially when the reliability of the scale-indicator is low). For the effect-coding method, second-order LGMs and their comparable first-order LGMs are relatively close in parameter estimates. When the latent variable at the first time point is standardized, the mean of the intercept-factor of the first-order LGM is close to 0 and not statistically significant; so is the mean of the intercept-factor of the second-order LGM through the effect-coding method, but those through two scaled-indicator methods are statistically significant and different from each other. According to our research results, the effect-coding method is recommended to scale and standardize the second-order LGMs, then comparable first-order LGMs are those based on the composite scores and their standardized models. For either the first-order or second-order LGM, the standardized results obtained by modeling composite total scores and composite mean scores are identical.
Subjects: Other Disciplines >> Synthetic discipline submitted time 2023-10-09 Cooperative journals: 《心理学报》
Abstract： Previous studies have found that participants benefit more from blocked learning in rule-based category learning but perform better with interleaved learning in information-integration category learning. In interleaved learning, participants need to generate four categories at the same time, which will create a high working memory load if applying a rule-based learning strategy and hence will encourage participants to switch from this sub-optimal strategy to information integration. However, previous studies always require passive conduct of blocked learning or interleaved learning. In real life, people will strategically switch between these two kinds of learning schedules. To grasp a better understanding, we compared passive and proactive learning schedules (blocked, interleaved, self-regulated, random). In addition, the categories used in previous studies are mutually exclusive, which contradicts real life where categories always overlapped each other and cannot be perfectly distinguished according to one or more combinations of features. For mutually exclusive structures, it is easy to confuse rule-based and information-integrated learners, and there is a countable difference in the learning speed of these two category structures. To gain more reliable results, an appropriate overlap level and the number of categories were chosen for this study. The classical four categories rule-based and information integration task is revised to contain overlapping stimuli. If classified by both two dimensions the highest accuracy was 90%. A 2 × 4 between-subject design was adopted. The dependent variables are accuracy and response time, and the first independent variable was the category structure: rule-based (RB) and information-integration (II). The second variable was the schedule of learning: blocked, interleaved, self-regulated, and random, with random presentation as the baseline condition. 265 college students were paid to participate in the experiment. Each participant should observe and report to which categories the line segment belonged. There were 100 trials each for both the learning phase and the test phase. Each phase comprised 25 trials for each category. For the test phase, a new set of stimuli are used and no feedback is provided. The behavioral data collected fit into a mathematical model to analyze what strategies participants used during tasks. The results showed a significant main effect of category structure. That is, the classification accuracy of the information-integration task is significantly higher than the rule-based task. The main effect of learning schedules was also significant. That is, the classification accuracy of interleaved, self-regulated, and random learning was significantly higher than that of blocked learning. Post hoc tests showed that the classification accuracy of the blocked learning was significantly lower than that of interleaved, self-regulated, and random learning under rule-based conditions. For the information-integration condition, the classification accuracy of the blocked learning was significantly lower than that of self-regulated. In addition, this study further analyzed learners' self-regulated learning behaviors under the overlapping category structure and found that for both rule-based tasks and information-integration tasks, learners' average length of blocked learning was significantly negatively correlated with their classification accuracy. A mathematical technique of the “Decision Bound Model” was used to analyze the data from the experiment. The results of model fitting showed that in both rule-based and information-integration tasks, self-regulated learners can use the optimal strategy more frequently. In conclusion, this study makes up for the deficiency of perfectly classified categories, finds the advantages of self-regulated learning and the disadvantages of blocked learning in category overlap, and preliminarily reveals the self-regulated learning advantages and information processing characteristics of overlapping category learning. It believes that category overlap interferes with the corresponding rules formed by learners for each category under the condition of blocking learning, which is not conducive to the blocked learning of rule-based tasks. In addition, the overlapping category structure will weaken the different information between categories and retain the common information within categories, which is not conducive to the interleaved learning of information-integration tasks. However, compared with passive learning, self-regulated learning has advantages in the learning of the two types of category structure because of its “decision-driven” and “data-driven” effects.
Subjects: Psychology >> Social Psychology submitted time 2023-03-27 Cooperative journals: 《心理学报》
Abstract： Doctors have to use different medical technologies to diagnose different kinds of illness effectively. Similarly, teachers have to use well designed tests to provide an accurate evaluation of students with different cognitive structures. To provide such an evaluation, we recommend to adopt the Cognitive Diagnostic Assessment (CDA). CDA could measure specific cognitive structures and processing skills of students so as to provide information about their cognitive strengths and weaknesses. In general, the typical design procedure of a CDA test is as follow: firstly, identify the target attributes and their hierarchical relationships; secondly, design a Q matrix (which characterizes the design of test construct and content); finally, construct test items. Within that designing framework, two forms of test are available: the traditional test and the computerized adaptive test (CAT). The former is a kind of test that has a fixed-structure for all participants with different cognitive structures, the latter is tailored to each participant’s cognitive structure. Researchers have not, however, considered the specific test design for different cognitive structures when using these two test forms. As a result, the traditional test requires more items to gain a precise evaluation of a group of participants with mixed cognitive structures, and a cognitive diagnosis computer adaptive test (CD-CAT) has low efficiency of the item bank usage due to the problems in assembling a particular item bank. The key to overcome these hurdles is to explore the appropriate design tailored for participants with different cognitive structures. As discussed above, a reasonable diagnosis test should be specific for the cognitive structure of target examinees so to perform classification precisely and efficiently. This is in line with CAT. In CAT, an ideal item bank serves as a cornerstone in achieving this purpose. In this regard, Reckase (2003, 2007 & 2010) came up with an approach named p-optimality in designing an optimal item bank. Inspired by the p-optimality and working according to the characteristics of CDA, we proposed a method to design the test for different cognitive structures. We conducted a Monte Carlo simulation study to explore the different test design modes for different cognitive structures under six attribute hierarchical structures (Linear, Convergent, Divergent, Unstructured, Independent and Mixture). The results show that: (1) the optimal test design modes for different cognitive structures are different under the same hierarchical structure in test length, initial exploration stage (Stage 0), accurately estimation stage (Stage 1); (2) the item bank for cognitive diagnosis computer adaptive test (CD-CAT) we built, according to the different cognitive structures’ optimal test design modes, has a superior performance on item pool usage than other commonly used item banks no matter whether the fixed-length test or the variable-length test is used. We provide suggestions for item bank assembling basing on results from these experiments.
Subjects: Psychology >> Social Psychology submitted time 2023-03-27 Cooperative journals: 《心理学报》
Abstract： Affirmation and negation are two main semantic and grammatical categories in any language. The propositional theory and experiential-simulations view were proposed to explain the processing mechanism of negative sentences. Both of their supporters have supplied plenty of empirical evidence, but neither of them can beat each other. Thus, the comprehensive theories, such as dual coding theory, LASS and symbolic interdependency hypothesis have been proposed to fill the gap. In the present study, we design two eye-tracking experiments to lend further support to comprehensive theories. In the two experiments, eye-tracking technical was adopted to explore the processing mechanism of negative sentences in different semantic contexts. In Experiment 1, the alternative choices (e.g., outspread arm) presented to the participants have close semantic connection with the negated events of the sentences (e.g., the arm isn’t crooked); In Experiment 2, the alternative choices (e.g., black skirt) presented to the participants have relatively weak semantic connection with the depicted negative events (e.g., the skirt isn’t blue). In summary, ‘blue-black’ has relatively weaker semantic connection than that of ‘outspread-crook’. In these two eye-tracking experiments, voice was used to present the negative sentence, and the corresponding pictures were presented at the moment of reading the words depicting the negated state (e.g., crook/blue). And the participants’ task was to choose which picture matched the sentence. The results demonstrated that, at the early stage of processing, there was no difference between the fixation probabilities to pictures depicting the negated state of affairs (crooked arm) and their alternative (outspread arm) in experiment 1 at time window 201~600 ms. In contrast, participants had higher fixation probabilities to pictures depicting the negated state of affairs (blue skirt) than that to pictures depicting the alternatives in experiment 2 at time window 401~600 ms. Then at the later stage, participants showed higher fixation probabilities to the pictures depicting alternatives to the pictures depicting the negated state of affairs from 601 ms in experiment 1 and 801ms in the experiment 2. Besides, the fixation probabilities to the pictures depicting the negated states were lower than the random level after 1001 ms in both of the two experiments. The results from the two experiments showed that, both propositional process and simulating process are necessary when processing negative sentences. Compared with processing negative sentence in weak semantic context, it’s easier for participants to get the actual state of event with the help of strong semantic context. In addition, participants will not keep the simulation of the negated state of event in his mind, which supports suppression hypothesis. In summary, the results support the symbolic interdependency hypothesis as well as suppression hypothesis.
Subjects: Psychology >> Social Psychology submitted time 2023-03-27 Cooperative journals: 《心理学报》
Abstract： Mathematically, a high-order factor model is nested within a bifactor model, and the two models are equivalent with a set of proportionality constraints of loadings. In applied studies, they are two alternative models. Using a true model with the proportional constraints to create simulation data (thus both the bifactor model and high-order factor model fitted the true model), Xu, Yu and Li (2017) studied structural coefficients based on bifactor models and high-order factor models by comparing the goodness of fit indexes and the relative bias of the structural coefficient in a simulation study. However, a bifactor model usually doesn’t satisfy the proportionality constraints, and it is very difficult to find a multidimensional construct that is well fitted by a bifactor model with the proportionality constraints. Hence their simulation results couldn’t extend to general situations.Using a true model with the proportionality constraints (thus both the bifactor model and high-order factor model fitted the true model) and a true model without the proportionality constraints (thus the bifactor model fitted the true model, whereas the high-order factor model fitted a misspecified model), this Monte Carlo study investigated structural coefficients based on bifactor models and high-order factor models for either a latent or manifest variable as the criterion. Experiment factors considered in the simulation design were: (a) the loadings on the general factor, (b) the loadings on the domain specific factors, (c) the magnitude of the structural coefficient, (d) sample size. When the true model without proportionality constraints, only factors (a), (c) and (d) were considered because the loadings on domain specific factors were fixed to different levels (0.4, 0.5, 0.6, 0.7) that assured the model does not satisfy the proportionality constraints.The main findings were as follows. (1) When the proportionality constraints were held, the high-order factor model was preferred, because it had smaller relative bias of the structural coefficient, and lower type Ⅰ error rates (but also lower statistical power, which was not a problem for a large sample). (2) When the proportionality constraints were not held, however, the bifactor model was better, because it had smaller relative bias of the structural coefficient, and higher statistical power (but also higher type Ⅰ error rates, which was not a problem for a large sample). (3) Bi-factor models fitted the simulation data better than high-order factor models in terms of fit indexes CFI, TLI, RMSEA, and SRMR whether the proportionality constraints were held or not. However, the bifactor models were less fitted according to information indexes (i.e., AIC, ABIC) when the proportionality constraints were held. (4) Whether the criterion was a manifest variable or a latent variable, the results were similar. However, for the manifest criterion variable, the relative bias of the structural coefficient was smaller.In conclusion, a high-order factor model could be the first choice to predict a criterion under the condition of proportionality constraints or well fitted for the sake of parsimony. Otherwise, a bifactor model is better for studying structural coefficients. The sample size should be large enough (e.g., 500+) no matter which model is employed.
Hits
YES
Hits
Downloads
Comment