The survey dataset of The Influence of theory of planned behaviour on purchase behaviour on social media

The research aims to study the correlations between attitudes of users, social norms, perceived behavioural control, and purchase behaviours. The research population consists of social media users in Malaysia. The data was collected from 205 respondents via a self-administered online survey. The theory of planned behaviour acts as the underlying theory in the research. Next, descriptive and hypothesis-testing quantitative analysis were adopted to probe the relationships between the variables. Moreover, G*Power was used to identify the minimum sample size, and SPSS v.22 was employed to examine the datasets.


Specification
Marketing Specific subject area Purchase Behaviour in Social Commerce Type of data

Value of the Data
• The data provide insights on the factors which influence the purchase behaviour of social media users in Malaysia. • The insights can provide suggestions to enhance the virtual and extended interaction between retailers and customers in the digitalised market. • External literature can reuse the data to probe the purchasing behaviour of different generations under the impact of social media influence. • The data can be used by academician to illustrate a course in statistical exploitation of survey data that focuses on structural equations modelling. • The research model can be adapted with moderation or moderating effect for researchers to explore the purchase behaviour of the younger generation in the respective countries.

Data Description
The raw data file and the survey questionnaire employed are provided alongside the data article as supplementary documents. The research model is built on the theory of planned behaviour shown in Fig. 1 , a social cognitive model used in social psychology to explain the com-  This is an individual resources and opportunities must, to some extent, determine the possibility of behavioural achievement.
PBC1: Frequency product advertisement on social media led me to buy. PBC2: I will choose social media as a reference channel for purchasing in the future. PBC3: I will recommend my friend choose social media as a reference channel for making purchase intention in the future. PBC4: I will recommend family choose social media as a reference channel for making purchase intention in the future.
This refers an individual willingness to conduct an action to buy product or services for consumption.
PB1: I am willing to buy a product promoted on social media. PB2: There is a high probability that I would purchase a product because of the impact of social media. PB3: I am easily influenced by advertisement on social media and further make a purchase behaviour. PB4: I had the experience of buying a product due to the influence of social media. [1 , 4] plexity of human behaviour where the behaviour intention is the direct antecedent used to forecast real conduct of a person willingness to put effort to conduct the behaviour concerned [1 , 3] . The empirical data are illustrated with 6 tables. Table 1 summarizes the constructs adopted from past research, with respective definitions and sources. Table 2 presents the demographic profile of the sample, categorized by gender, age, ethnicity, income level, and occupation. Table 2 also includes the behavioural background of participants, in terms of their experiences with social media. Table 3 displays the mean, standard deviation, and reliability of the measured constructs. The reliability of the data is probed using Cronbach's alpha, as shown in Table 3 . The Cronbach's alpha score is above the minimum threshold of 0.7 (ATTD = 0.922; SN = 0.919; PBC = 0.934; PB = 0.912), hence the data is reliable and consistent [7 , 8 , 14 , 21 , 22] . Table 4 shows the model summary of the coefficient of determination (R 2 ) model. The R -value (0.918) and Rsquare (0.842) in Table 4 demonstrated that the effectiveness of the model in determining the dependant variables is substantial, as based on the recommended explained variance (R 2 ) values by Hair et al. [5] , R 2 of 0.75, 0.50, and 0.25 signify substantial, moderate, and weak predictive power [9 , 15 , 17 , 19] . Table 5 shows the readings of ANOVA. Finally, the model coefficients, which include the Beta and standard error of the measured constructs are presented in Table 6 . The    findings from Tables 5 and 6 indicate significant relationships between all the constructs, by which attitudes of users, social norms, and perceived behaviour control are positively related to the purchase behaviour of social media users in Malaysia; and can be used to predict the purchase behaviour of social media users in Malaysia.

Experimental Design, Materials and Methods
The research employed quantitative methods with the deployment of self-administered online survey questionnaires as the measurement tool [16 , 17 , 18 , 20] . Moreover, the survey questionnaires consist of 16 measurement items for the 4 latent constructs probed. The measurement items were adopted and adapted from past literature [4 , 11] and anchored on a 5-point Likert scale which ranged from 1: "Strongly Disagree" to 5: "Strongly Agree". 340 questionnaires were distributed with a non-probability, convenience sampling method via Google Forms to the individual who uses social media, howbeit only 219 filled-pout responses were collected. Snowball sampling was used where the survey link is shared via WeChat Moments for voluntary participations. No incentives were given for the survey. Of the 219 empirical data collected, only 205 responses were qualified and eligible for analysis about data filtering, the remaining respondents were eliminated due to inexperience with social media. The filtered dataset still exceeds the minimum sample size determined using G * Power 3 with a statistical power of 0.8, margin error of 0.05, and effect size of 0.15, with 3 predictors was used to determine the minimum sample size, as recommended by past literature [10 , 16 , 18] . Besides, in reference to Roscoe et al. [12] rule of thumb, a sample size between 30 and 500 should be sufficient for most studies [6 , 13] . SPSS v.22 was employed to conduct the data analysis, particularly the Cronbach's alpha to examine the reliability and validity [2] , then structure equation modelling to estimate the coefficients of the latent variable, which is believed to be the appropriate estimation methods for multiple regression analysis [8 , 23 , 24] .

Ethical Statements
The self-administered survey that is non-experimental in nature was conducted under complete anonymity for the participants. No personal or sensitive information that can be used to identify the respondents were collected. Besides, the consent of the respondents to partake in