Essay On Demographic Differences And Perceptions Of Self-Checkout Counters In Singapore

The main objective of the research study was to investigate how users and non-users in self-checkout counters differed in the retail industry in the country of Singapore. Differing from the previous research studies in this topic which only investigated the usage or avoidance of self - checkout counters based on an individual’s avoidance to use of technology, the authors aim to find through this research if there were any situational reason that affected a customer’s usage or avoidance of self – checkout counter. The main research questions addressed in the article by the researchers were –

RQ1: Are there demographic differences between users and non-users of self-checkout counters in Singapore?

This research question was adapted by the researchers to compare the users and non-users based on their demographic features. The demographic information acquired from the survey participants during the survey included – age, gender, level of education and whether the participant used self – checkout counter or not. This investigation would help understand the researchers if there is any pattern in using self – checkout counters by customers in a shopping mall based on the acquired demographic information.

RQ2: Are there differences between users and non-users of self-checkout counters in how they perceive the characteristics of self-checkout counters in Singapore (i.e., relative advantage, complexity, reliability, compatibility and fun)?

This question was adapted by researchers to examine the difference in perceived advantage of using self - checkout counters between users and non - users. The perceived advantage was assessed in terms of compatibility, reliability, complexity, fun and relative advantage.

RQ3: Are there differences between users and non - users of self - checkout counters in how situational factors affect their intention to use the terminals?

This question was developed for the purpose of investigating the situational factor effectiveness on decision for using self - checkout counters. Through this investigation the researchers aim to inform if any specific situation drives a customer to use or avoid using a self – checkout counter at any shopping mall.

A) The survey instrument comprised of three sections. The first section was used to determine if the participant was user of self-checkout counter. The second section was used to measure the perceived advantage of using self - checkout counters on a five point Likert scale. The third section was used to measure the likelihood of participants of using a self-checkout counter under different situational circumstances. The first section was named as Respondents’ profile. The second section was named as evaluation of self - checkout counters. The third section was named as situational factors in the decision to use self - checkout counters.

B) The second section of the survey instrument measured the perceived advantage of using self - checkout counters by participants. It was subdivided into four factors - perceived complexity, relative advantage, fun and perceived reliability of using self – checkout counter while shopping at any shopping mall. Each dimension or factor was subdivided and measured with a three item scale. The section also measured how much compatible were the self - checkout counters with the lifestyle of the participant.

Data Collection

C) Cronbach’s alpha is used to quantify the internal consistency of a scale. It shows how reliable a scale is (Vaske, Beaman and Sponarski 2017). Researchers use Cronbach’s alpha to investigate the internal consistency of a survey questionnaire composed of more than one Likert-scale items (Taber 2018). In general, a Cronbach’s alpha score greater than 0.7 is considered to be reliable (Chan and Idris 2017). The researchers should indicate how reliable each scale is in a survey analysis which is why the Cronbach’s alpha coefficients must be reported in the paper. According the authors the Cronbach’s alpha values for relative advantage was 0.85 which indicates the scale was reliable. For perceived complexity the Cronbach’s alpha score was 0.84 and for the scales reliability and fun the values were 0.72 and 0.92 respectively.

Convenience sampling method was implemented in the procedure for collecting data. Convenience sampling is a type of sampling where the researcher uses a sample that is convenient to the purpose of the research. It is a type of non - probability sampling (Vehovar, Toepoel and Steinmetz 2016). Non – probability sampling refers to a sampling procedure in which the members of the population do not all have equal chances of getting included in the sample. The researcher collected data from a suburban resident in the West of Singapore. This was convenient for the researcher as the locality had three shopping malls and a subway station. This helped researcher to collect data from many pedestrian travellers who visited any of the shopping malls. The response by each participant in the survey was independent of one another.

A) The authors conducted statistical quantitative analysis to examine whether there was any difference between users and non - users in terms of demographic characteristics. The statistical test used was chi - squared test for independence. Chi – squared test of independence is used to check whether the observed frequency in one or more categories of a data is significantly different from the expected frequencies of those categories. The null hypothesis in chi – squared test of independence states that the observed frequency has no statistically significant difference from the expected frequencies in a contingency table. The alternate hypothesis states that observed frequency has a statistically significant difference from the expected frequencies in a contingency table. If the observed frequencies have a significant difference from the expected frequencies then it is indicated that there is a relationship between the dependent and independent variable. The authors reported the chi - squared statistics that supported their evidence. For example: According to the authors there was no association between gender and use of self-checkout counters . The p - value associated with the test statistic was 0.25 which was greater than the level of significance.

B) According to the authors, there was no association between age and usage of self-checkout counters . “4” represents the degrees of freedom for the test. The variable age was divided into five categories. The degree of freedom is obtained by as (Number of rows - 1) × (Number of columns – 1) (Rana and Singhal 2015)

Statistical Tests

In this case,

(Rows -1) × (Columns - 1) = (5 - 1) × (2 - 1) = 4

Where, number of rows = 5

And number of columns = 2 (User and Non-user).

The independent samples t - test were conducted by researchers to investigate the significance in mean difference of perceived advantage of using self - checkout counter in terms of perceived complexity, relative advantage, fun and reliability between users and non-users. According to the authors, due to a large sample size, even though the mean differences were statistically significant the effect sizes were small and moderate. When the sample size is very large, the mean difference is almost always observed to be significant. The effect size is however not affected by the sample size.

According to the authors, a recent study indicated that people shopping in crowded places tend to perceive a pressure of time and try to shop quickly. In such a circumstance, shoppers tend to avoid using self-checkout counters. Research could be conducted to test whether shopping under crowded situations in public has any impact on the usage of self-checkout counters. Data could be collected regarding likelihood of shoppers of using self-checkout counters for two independent sample of people one being shoppers who shopped under crowded circumstances and the other being a sample of shoppers who did not shop under crowded situations through Likert Scale response. A t - test could be performed to test the significance in mean difference of perceived likelihood of using self - checkout counter between shoppers who shopped under crowded situations and those who shopped under non crowded situations.

1. Based on the given data the following frequency table was developed for “Take Up”

Table 1: Frequency table for Take Up

2. The null and alternate hypothesis to test for RQ1 are as follows –

H₀: Proportion of people in favour of purchasing from Amazon Fresh is not different from 0.5.

H₁: Proportion of people in favour of purchasing from Amazon Fresh is greater than 0.5.

A binomial test revealed that the proportion of people in favour of shopping at Amazon Fresh was significantly greater than 0.5 (p = .009). The lower 95 % confidence for proportion of people in favour of shopping at Amazon Fresh was 0.539 which was greater than 0.5.

3. The null and alternate hypothesis to test for RQ2 are as follows –

H₀: Proportion of people in favour of shopping at Amazon Fresh for those who find using mobile technology comfortable differ significantly from those who did not find using mobile technology comfortable.

H₁: Proportion of people in favour of purchasing from Amazon Fresh for those who find using mobile technology comfortable differ significantly from those who did not find using mobile technology comfortable.

A z - test for difference in two proportions indicated that the proportion of people in in favour of making purchase from Amazon Fresh for those who find using mobile technology comfortable differ significantly from those who did not find using mobile technology comfortable (z = 2.26, p = .024).

Hypothesis test was conducted to investigate whether the difference in observed point estimates of sample proportions from the hypothesized proportion is due to sampling error or actually significant across the population (Emmert-Streib and Dehmer 2019). By simply comparing the sample proportions it is possible that one makes wrong conclusions about the actual population (Park 2015). The hypothesis testing is an inferential statistical approach through which one can obtain conclusions about the complete population based on sample statistics under a certain specified level of significance.

Table 2: Descriptive Summary for entrepreneurship intention

A Levene’s test was conducted to verify whether the variance in entrepreneurship intention score between the two groups of students one with entrepreneurship module and the other without entrepreneurship module was same or statistically significantly different.

Levene’s test for homogeneity of variances revealed that the variance in entrepreneurship intention of students with no entrepreneurship module was not different from that of students with entrepreneurship module, F (1, 132) = 1.039, p = 0.31.

Shapiro – Wilk test was conducted to test if the entrepreneurship intention score in the sample was normally distributed.

A Shapiro - Wilk test for normality revealed that the entrepreneurship intention scores were normally distributed (SW = 0.988, p = 0.324).

H₀: The average entrepreneurship intention score of students with no entrepreneurship module does not differ from the average score of students with entrepreneurship module.

H₁: The average entrepreneurship intention score of students with no entrepreneurship module differed significantly from the average score of students with entrepreneurship module.

As the assumptions for normality and homogeneity of variances in entrepreneurship intention score between the two groups were met we proceed to test the hypothesis with a parametric test (Kim 2015). Independent sample t - test was performed to investigate the significance in mean difference between students with no entrepreneurship module and students with entrepreneurship module (Mishra et al. 2019).

An independent samples t - test revealed that the mean difference in the entrepreneurship intention score between students with no entrepreneurship and students with entrepreneurship module was statistically significant, t (132) = - 2.640, p = 0.009, mean difference (95% CI) = - 1.313 (- 0.807, - 0.106). Effect size was measured through Cohen’s d which was – 0.459 indicating a very less or no effect. When the effect size Cohen’s D value is less than 0, it is indicated that there was no effect of the variation of the independent variable on the dependent variable.  

H₀: The average entrepreneurship intention score for all departments in the university is the same.

H₁: The average entrepreneurship intention score in at least one department of the university is significantly different from the others.

One - way ANOVA was conducted to investigate the significance in mean difference in entrepreneurship intention score among the various departments in the university (Ross and Willson 2017). ANOVA is performed to investigate the mean difference between more than one groups. One – way ANOVA is performed when there is only one independent variable. The independent variable in this case was university department and since there were more than two departments in the university a one - way ANOVA was performed.

A one-way ANOVA revealed that the mean entrepreneurship intention score in at least one department of the university is significantly different from the others, F (2, 131) = 7.738, p < 0.001, η2 = 0.106, = 0.091. The effect size was moderate. η2 value greater than 0.06 and less than 0.14 is considered to indicate moderate effect. In the present case scenario, the η2 value being 0.106 indicated that the effect of department on entrepreneurship score was moderate. The statistic was 0.09 which also indicated a moderate effect of the independent variable university department on the dependent variable entrepreneurship score.

In hypothesis test the p - value is obtained to check the significance of the test statistic. In hypothesis testing, the p - value helps the researcher in deciding the statistical significance of observed effect with respect to the null hypothesis (Krueger and Heck 2019). The p - value provides a basis to the researcher to decide whether to reject a null hypothesis or to not reject it (Ioannidis 2018). If the p - value is less than the decided level of significance for the hypothesis test then the null hypothesis in a hypothesis is always rejected. The p – value indicates the extent to which a null hypothesis in any hypothesis test is supported by the sample data. Higher p – values always indicate that the data in a sample favour or supports the null hypothesis. Smaller p – values always indicate the data in the sample does not support the null hypothesis. The null hypothesis always states that there is no relationship between a dependent and the independent variable. A limitation of the p - value is that it does not give the probability that the null hypothesis in a hypothesis test is true, or the probability that the data were produced because of chance at random (Solla et al. 2018). For large sample size the p – value is always very less, which indicates statistically significant relationship between dependent and independent variable, even though the effect size of the relationship could be moderate or small. Thus, p – value does not indicate the effect size of the association between between the independent variable and the dependent variable (Lee 2016). This is the reason why in any hypothesis test, the effect size is also computed and reported along with the p – value, especially when the sample size is considerably large.

References

Chan, L.L. and Idris, N., 2017. Validity and reliability of the instrument using exploratory factor analysis and Cronbach’s alpha. International Journal of Academic Research in Business and Social Sciences, 7(10), pp.400-410.

Emmert-Streib, F. and Dehmer, M., 2019. Understanding statistical hypothesis testing: The logic of statistical inference. Machine Learning and Knowledge Extraction, 1(3), pp.945-961.

Ioannidis, J.P., 2018. The proposal to lower P value thresholds to. 005. Jama, 319(14), pp.1429-1430.

Kim, T.K., 2015. T test as a parametric statistic. Korean journal of anesthesiology, 68(6), p.540.

Krueger, J.I. and Heck, P.R., 2019. Putting the p-value in its place. The American Statistician, 73(sup1), pp.122-128.

Lee, D.K., 2016. Alternatives to P value: confidence interval and effect size. Korean journal of anesthesiology, 69(6), p.555.

Mishra, P., Singh, U., Pandey, C.M., Mishra, P. and Pandey, G., 2019. Application of student's t-test, analysis of variance, and covariance. Annals of cardiac anaesthesia, 22(4), p.407.

Park, H.M., 2015. Hypothesis testing and statistical power of a test.

Rana, R. and Singhal, R., 2015. Chi-square test and its application in hypothesis testing. Journal of the Practice of Cardiovascular Sciences, 1(1), p.69.

Ross, A. and Willson, V.L., 2017. One-way anova. In Basic and advanced statistical tests (pp. 21-24). Sense Publishers, Rotterdam.

Solla, F., Tran, A., Bertoncelli, D., Musoff, C. and Bertoncelli, C.M., 2018. Why a P-value is not enough. Clinical spine surgery, 31(9), pp.385-388.

Taber, K.S., 2018. The use of Cronbach’s alpha when developing and reporting research instruments in science education. Research in science education, 48(6), pp.1273-1296.

Vaske, J.J., Beaman, J. and Sponarski, C.C., 2017. Rethinking internal consistency in Cronbach's alpha. Leisure sciences, 39(2), pp.163-173.

Vehovar, V., Toepoel, V. and Steinmetz, S., 2016. Non-probability sampling. The Sage handbook of survey methods, 1, pp.329-45.