Big Data Analytics and Sales Performance

Materials and Methods

Measures

This research used a structured questionnaire for data collection that was divided into two parts. The first part was about the respondents' demographic information, and the second part contained items of the variables. For content validity, the research adopted measures from prior studies of a similar context. Table 2 presents the items of the questionnaire and the sources of these items. Moreover, a seven-point Likert scale (1 = strongly disagree to 7 = strongly agree) was adopted to numerically measure the items because the seven-point Likert scale is one of the most widely used scales.

Pilot Test

Instrument validity is an important issue to be addressed before entering the data collection phase. Pretest and pilot test techniques were adopted to guarantee questionnaire validity. In the pretesting phase, a group of two full professors and three doctoral students with expertise on instrument development were consulted. The questionnaire items were discussed with the experts, and the proper wording, formation, appropriateness, and length of the items were ensured. After pretesting, 40 respondents were chosen from the selected sample size of the research. These respondents were not included in the main research analysis to avoid subsequent behavior. Table 3 presents the results of the pilot testing. The result shows that the loadings of the BCS ranged from 0.936 to 0.948, with a Cronbach's alpha score of 0.974. The factor loadings of personalization ranged from 0.817 to 0.875, with a Cronbach's alpha score of 0.958. The loading ranges for AA and IRK were 0.813–0.912 and 0.877–0.896, respectively, with Cronbach's alpha scores of 0.940 and 0.987, respectively. The loading ranges for CIMC, CRUC, and CWBC were 0.803–0.910, 0.776–0.943, and 0.804–0.945, respectively, and the Cronbach's alpha scores were 0.847, 0.944, and 0.950, respectively. The factor loadings of PSP ranged from 0.919 to 0.935, with a Cronbach's alpha score of 0.959. The values of the factor loadings and Cronbach's alpha for all the constructs exceeded the threshold point of 0.7, which ensured the validity of the instrument.

Sampling and Data Collection

The pharmaceutical organizations of Pakistan were selected as the population for this study, and the survey method was adopted for data collection from the targeted sample to test the research model. The survey method is suitable for analyzing the behavioral relationship among variables. The contact data of the pharmaceutical organizations were obtained from the Drug Regulatory Authority of Pakistan, a regulatory body that controls the pharmaceutical organizations of Pakistan. The convenience sampling technique was used to select the sample size as it is suitable for cases in which the availability of respondents varies. First, the human resource departments (HRDs) of the pharmaceutical organizations were contacted and provided with the purpose and brief introduction of the research, as well as an assurance that the data collected from the employees would only be used for research purposes and that the research findings would be shared with them. After securing permission and with the help of the HRDs, data were collected through an online survey from the employees of sales and marketing departments linked with the usage of different sales management systems. An online survey is suitable because of consistent data collection. The G∗Power software was used to predict the suitable sample size as it can ensure sample adequacy before and after data collection. The required sample size was 262 at "A priori: Compute the required sample size". Therefore, 750 questionnaires were distributed, and 441 of them were returned with responses. A total of 416 valid responses were selected for final analyses after discarding biased responses. Meanwhile, "Post hoc: Compute achieved power" was applied to the collected sample size of 416, and the power was determined to be 0.995, indicating the adequate strength of the study sample.