Unlocking insights: A comprehensive dataset analysis on the acceptance of computational thinking skills among undergraduate university students through the lens of extended technology acceptance model, HTMT, covariance-based SEM, and SmartPLS

Abstract

In light of the increasing importance digital economy, the significance of computational thinking has grown exponentially, becoming imperative in both workplace and academic settings such as universities. This article addresses the critical need to comprehend the factors influencing the acceptance of computational thinking. The dataset introduces an extensive questionnaire comprising five constructs and 25 items, rooted in the extended Technology Acceptance Model. Notably, the model incorporates facilitating conditions and subjective norm, providing a comprehensive framework for understanding acceptance. Data collection involved 132 undergraduate university students sampled through purposive sampling, specifically targeting courses with a focus on computational thinking. The resulting dataset serves as a valuable resource for future research, offering detailed insights into the factors determining the acceptance of technology in educational contexts beyond mere thinking skills. Given the scarcity of research on technology acceptance in developing nations, this dataset holds particular significance, serving as a foundation for potential cross-cultural comparisons. The dataset contributes to the field by presenting a robust acceptance model, explaining 74.2 per cent of the variance in behavioural intention, 60.2 per cent in perceived usefulness, and 56.1 per cent in perceived ease of use. This high explanatory power positions the dataset as a superior resource for replication, benchmarking, and broader applicability in diverse contexts, thereby enhancing the understanding of computational thinking acceptance across different populations and settings. This dataset stands among the pioneering efforts to assess the novel covariance-based structural equation model algorithm within SmartPLS 4, presenting a valuable resource for future research employing the same mechanism.

Specifications Table

Subject	Education
Specific subject area	Technology acceptance, learning behaviour, computational thinking skills, applied statistics, social sciences.
Data format	Raw data, Table, Figure, Questionnaire (constructs and items)
Type of data	Table, Image, Chart, Graph, Figure, Questionnaire items
Data collection	The survey results regarding the acceptance of computational thinking skills among undergraduates provided the data. The survey items were derived from well-established academic papers about the Technology Acceptance Model, subjective norm, and facilitating condition. Purposive sampling was employed to disseminate the questionnaire to university students across three campuses of Malaysian universities. The sample was specifically targeted towards individuals who possessed an academic background in computational thinking. The participants were duly informed of the research and publication objectives of the study and were requested to indicate their agreement to participate in the survey. Additionally, they were assured that their responses would be kept anonymous.
Data source location	A survey was conducted on three university campuses situated in the southern and central regions of Malaysia in order to collect the data. The universities that participate remain anonymous.
Data accessibility	Repository name: Mendeley Data Data identification number: 10.17632/552zh82tky.1 Repository name: OSF Data identification number: 10.17605/OSF.IO/TXBQR

1. Value of the Data

• •The dataset comprises data on the principal antecedents that influence the behavioral intention of university students to accept and embrace novel technology implementation. This information is not restricted to cognitive abilities alone but is applicable to other technical domains as well. The employment of the dataset's results has the potential to enhance the collective comprehension of university students' technological adoption. These students not only represent the workforce of the future but also establish trendsetters for the workplace of the future.
• •The data supply other researchers with a knowledge of the endogenous and exogenous factors that play a direct impact on university students' acceptance and use of technology. The model derived from the data demonstrates an explanatory power of 74.2 per cent for behavioral intention, 60.2 per cent for perceived usefulness, and 56.1 per cent for perceived ease of use. This indicates that the model is capable of elucidating a significant portion of the variability observed in the three endogenous variables being investigated. Therefore, it may function as a significant initial concept for subsequent researchers.
• •The data could potentially provide universities with a strategic resource for identifying critical determinants that impact the acceptance or adoption of technological innovations by university students. It may function as the cornerstone for subsequent investigations by researchers and universities to guarantee a successful integration of innovative advancements, such as cutting-edge instructional technology or inventive cognitive skills.
• •The dataset may be reused by researchers for cross-cultural comparisons. As in the past, research on university students' adoption of technology has mostly concentrated in the global north, resulting in insufficient understanding of the situation in developing nations.
• •The dataset was analyzed with the SmartPLS 4 software's new covariance-based structural equation model algorithm. This information will serve as a valuable base for future studies that intend to adopt the same data analysis mechanism.

2. Background

The dataset utilized in this research aims to explore the acceptance of computational thinking among undergraduate university students, comprising 132 samples. The foundational framework for this study is the Technology Acceptance Model [1], which has been expanded by incorporating facilitating conditions and subjective norms as two external variables as formulated in the theoretical framework illustrated in Fig. 1. The Technology Acceptance Model is used not only to understand the acceptance of various technologies, but it is also being applied to study critical skills in an educational setting. Such skills may include online teaching [2], creativity [3], and study skills [4], and the model seems to be well-suited for comprehending computational thinking skills as well. The analytical approach employs a covariance-based structural equation model, encompassing three key phases: Confirmatory Factor Analysis, the construction of a measurement model, and the synthesis of a structural model. Convergent validity is assessed through measures such as factor loading, composite reliability, and the Average Variance Extracted (AVE). Discriminant validity is verified using the Heterotrait-Monotrait (HTMT) ratio [5], representing a novel method for establishing discriminant validity [6]. Additionally, the dataset undergoes analysis using SmartPLS 4.0.9.9 [7], marking a departure from its historical association with the partial-least square method to accommodate a covariance-based approach. This study attests to the seamless integration of SmartPLS upgrades, affirming its compatibility with covariance-based methods in real-world data applications. The adoption of the covariance-based structural equation model presents a promising avenue for future studies employing structural equation modelling analyses.

Fig. 1.

Theoretical framework.

3. Data Description

The dataset is available as a supplemental file using Mendeley Data and the data were collected through a correlational study deploying a questionnaire to study the degree of acceptance of computational thinking among undergraduate students. The data collection took place across three university campuses in Malaysia by the samples who took the questionnaire online mostly using their smartphone or table.

The questionnaire was designed to understand the level of acceptance of computational thinking skills among undergraduate university students. The computational thinking is a critical set of skills that is fundamental not only to computer scientists but also to everyone as put forward by Wing [8]. The significance of computational thinking has expanded beyond the domain of computer science throughout time, as seen by the research conducted by Cai et al. [9] as well as Zhou and Tsai [10].

The dataset is comprised of a spreadsheet-formatted data file with 132 columns and 28 rows of data. The 132 columns in the table indicate the 132 samples that were included in the research. The 28 rows provide information regarding the samples' responses to the demographic data query and the specific items (n = 25) of each construct. The report presents a set of three demographic factors: age, gender (with males denoted by the code 1 and females by the code 2), and year of study (which is coded according to year). The data set comprises five constructs: subjective norm, facilitating condition, and behavioral intention to accept computational thinking. Perceived usefulness and perceived ease of use are the other constructs.

The instrument employs a five-point Likert scale, where 5 indicates “strongly agree” and 1 indicates “strongly disagree.” The sample profile comprises 91 females (69 %) and 41 males (31 %). In relation to the age distribution of the samples, the proportion of individuals aged 23 comprises the largest proportion of the samples, including 38.5 per cent (n = 50) of the total samples. Subsequently, samples aged 25 and above comprised 40 samples, followed by samples aged 22 at 25 samples, eight samples at 24, 4 samples aged 21 years old, and lastly 3 samples aged 20 years old. The majority of the samples are second-year students (n = 70), followed by third-year students (n = 41), and a small portion of final year students (n = 14) and first year students (n = 7). Table 1 provides the demographic information of the samples based on frequency and percentage.

Table 1.

Demographic information.

	Demography	Frequency, f	Percentage, %
Gender	Male	41	31
	Female	91	69
Age	20 years old	3	2.5
	21 years old	4	3
	22 years old	25	19
	23 years old	50	38.5
	24 years old	8	6
	25 years old & above	40	31
Year of Study	1st year	7	5
	2nd year	70	53
	3rd year	41	31
	4th year	14	11

After the data collection, the data and the instrument were tested for its convergent validity using the Confirmatory Factor Analysis technique to measure the factor loading (λ), Average Variance Extracted (AVE), and Composite Reliability (CR). The output of the convergent validity is shown in Table 2.

Table 2.

Convergent validity based on factor loading, Cronbach's alpha, AVE, and CR.

Constructs	Items	Factor Loading, λ	Cronbach's alpha, α	AVE	CR
Behavioral Intention	BI1	0.928	0.968	0.834	0.968
	BI2	0.921
	BI3	0.928
	BI4	0.912
	BI5	0.872
	BI6	0.916
Perceived Usefulness	PU1	0.893	0.973	0.860	0.973
	PU2	0.953
	PU3	0.957
	PU4	0.950
	PU5	0.910
	PU6	0.900
Perceived Ease of Use	PEU1	0.898	0.965	0.822	0.965
	PEU2	0.892
	PEU3	0.915
	PEU4	0.926
	PEU5	0.919
	PEU6	0.889
Facilitating Condition	FC1	0.844	0.929	0.820	0.933
	FC2	0.936
	FC3	0.932
Subjective Norm	SN1	0.915	0.921	0.751	0.919
	SN2	0.949
	SN3	0.794
	SN4	0.796

Table 2 shows that the dataset, measurement model, and instrument posed an excellent degree of convergent validity based on the threshold value of 0.5 sets for the standardized factor loading, λ according to the popular literature by Hair et al. [11]. The values of AVE are all beyond the threshold value of 0.5 [12] and the values of CR exceed the threshold value of 0.7 [11]. This serves as a warrant that the collected data, the generated dataset, the measurement model and the instrument used as valid for understanding the constructs designated. The notion is supported by the consistency of the dataset as indicated by Cronbach's alpha values of each construct that range from 0.921 to 0.973.

However, another issue with using questionnaires for the purpose of research is the multicollinearity of the constructs involved in the study. As each construct represents a variable that being included in the study, it is highly likely that one construct might be heavily influenced by another construct that jeopardizes the findings of the study. Thus, it is highly imperative to ensure the minimum likelihood of multicollinearity using discriminant validity measurement. In study using covariance-based structural equation model, the most common method of measuring discriminant validity is through the calculation of the Fornell-Larcker Criterion as established by Fornell and Larcker [12]. It is previously widely applied in studies about Technology Acceptance Model such as in studies by Wang et al. [13] and Almarzouqi et al. [14]. Yet, a newer and comprehensive method of detecting the establishment of discriminant validity using the Heterotrait-Monotrait (HTMT) ratio is put forward by Henseler et al. [5] for primary usage of partial-least square structural equation model has been proven applicable for covariance-based structural model by Rosli and Saleh [6]. The dataset of this study is therefore tested using HTMT as shown in Table 3.

Table 3.

Discriminant validity using HTMT.

	BI	FC	PEU	PU	SN
BI
FC	0.594
PEU	0.803	0.638
PU	0.824	0.518	0.776
SN	0.719	0.775	0.768	0.645

The values of the HTMT ratio of the constructs are excellent with the highest one being between PEU and PU at 0.776. However, this is not uncommon, given that the strong correlation between the two variables is suggested by the Technology Acceptance Model in the seminal work by Davis [1]. Yet, the value is far below the threshold value set at 0.90 [5] to conclude that discriminant validity has been established [15]. As all the values of the HTMT ratio are well below 0.90, the dataset and the instrument used exhibit good discriminant validity.

The establishment of convergent and discriminant validity shows that the dataset and the instrument are capable of drawing an accurate conclusion on the acceptance of computational thinking based on the Technology Acceptance Model. The analysis progresses to the establishment of a structural model by integrating facilitating condition and subjective norm into the Technology Acceptance Model as precursors to perceived usefulness and perceived ease of use in the technology acceptance model. The structural model is as in Fig. 2.

Fig. 2.

Structural model.

The model quality of the structural model is as in Table 4.

Table 4.

Model quality of the structural model.

Indices	Model Fit
Chi-Square, χ2	620.069
Degree of freedom, df	267.000
χ2/df	2.322
Comparative Fit Index, CFI	0.922
Tucker-Lewis Index, TLI	0.913
Root Mean Square Error of Approximation, RMSEA	0.100
Standardized Root Mean Square Residual, SRMR	0.051

To understand the relationships between the variables in the extended Technology Acceptance Model, the following path analysis on the structural model was performed as in Table 5.

Table 5.

Path analysis of the structural model.

Path	Path coefficients, β	Parameter estimates, β	Standard errors	T values	p values
FC to PU	0.008	0.008	0.094	0.084	0.933
FC to PEU	0.134	0.144	0.111	1.296	0.197
SN to PU	0.085	0.089	0.121	0.738	0.462
SN to PEU	0.644	0.733	0.123	5.933	0.000
PEU to PU	0.705	0.647	0.094	6.867	0.000
PU to BI	0.468	0.499	0.089	5.578	0.000
PEU to BI	0.447	0.437	0.083	5.279	0.000

Tables 2 to 5 shows the inferential statistics of the data analysis based on the covariance-based structural equation model.

4. Experimental Design, Materials and Methods

The study uses correlational research designed to understand the acceptance of computational thinking skills among undergraduate university students. The dataset was collected via the questionnaire that was implemented as the data collection instrument. Twenty-eight items were designed to evaluate five constructs, comprised of two exogenous factors and three endogenous variables in addition to three demographic variables. The items to measure three variables from the Technology Acceptance Model were derived from previous study by the founder of the model [1].

The Technology Acceptance Model constructs consisted of perceived usefulness, perceived ease of use, and behavioral intention constructs; each construct was measured using six items, for a total of eighteen items used to measure the Technology Acceptance Model constructs [1,6,16]. The items used to measure the facilitating condition were adapted from Thompson et al. [17], while the items employed to assess subjective norm were adapted from Taylor and Todd [18]. Three and four items were used to measure the two constructs, respectively.

The responses to the questionnaire were subsequently exported to Google Sheets and saved as a Microsoft Excel file after being disseminated over the online platform Google Forms. After encoding the questionnaire questions into coded items, the data was processed into a processed dataset prior to being transferred into SmartPLS 4 [7]. Subsequently, the data were screened to ensure they adhered to the multivariate normality principle. This is accomplished by limiting the skewness to a range of -3 to +3 and the kurtosis to -10 to +10. Such conditions are met to qualify the data for analysis using the covariance-based structural equation model method. The processed dataset was further analyzed with the SmartPLS 4 software using its covariance-based structural equation model algorithm.

The data were collected from three university campuses in the southern and central parts of Malaysia. The universities were not named for the purpose of anonymous reporting of the universities involved. The universities are sampled using purposive sampling techniques due to the fact that they offer undergraduate courses that have computational thinking skills as the core skills needed to succeed in the courses offered. The sampling technique of the samples is also using purposive sampling techniques as it enables the researchers to tap into the undergraduate students with experience of engaging with computational thinking skills. A total of 132 responses were received by the researchers and the data collection took place from April to December 2023. The methodology process is illustrated in Fig. 3.

Fig. 3.

Methodology process.

Limitations

The dataset exhibits certain limitations; however, these do not compromise the overall quality of the data. The sample size, consisting of 132 participants, reflects the intentional use of purposive sampling techniques to specifically target university students enrolled in courses emphasizing computational thinking skills. Although the limited sample size is acknowledged, it is a deliberate outcome of the chosen sampling technique. Notably, a portion of the variance remains unexplained in the dataset, with figures of 25.8% for behavioral intention, 39.8% for perceived usefulness, and 43.9% for perceived ease of use. Despite these unexplained variances, the dataset's explanatory power surpasses that of the unknown variances.

Ethics Statement

The participants provided informed consent to participate in the study before starting the questionnaire. They were also informed of the research's objectives, which encompassed academic research and publishing, and were assured of the confidentiality of the report. The dataset does not include any personally identifiable information, including the names of the universities and courses that were included. There is no ethical approval necessary for this research.

CRediT authorship contribution statement

Mohd Shafie Rosli: Conceptualization, Data curation, Formal analysis, Visualization, Methodology, Writing – original draft, Supervision, Funding acquisition. Muhammad Fairuz Nizam Awalludin: Validation, Investigation. Cheong Tau Han: Writing – review & editing, Investigation, Funding acquisition. Nor Shela Saleh: Methodology, Writing – review & editing, Validation. Harrinni Md Noor: Funding acquisition.

Data Availability

• Dataset (Original data) (Mendeley Data)

• Questionnaire - Constructs and Items (Original data) (OSF).