Observations

#1. Applicability of TAM

Factor Analysis

To reduce these 18 items into a fewer number of factors that would explain the majority of variance of these items, factor analysis was conducted on the data.

The suitability of factor analysis was determined by two criteria viz. Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy as a measure of homogeneity of variables. A KMO measure of above 0.60 is acceptable for factor analysis and Bartlett's test of sphericity tests whether the correlation matrix is an identity matrix, which would indicate that the factor model is inappropriate. The corresponding Chi-Square value is 1.210E3, which is significant even at 1% significance level. Hence factor analysis is suitable for our data set.

Principal Component Analysis was conducted on the data, which yielded a six-factor solution, with eigenvalues greater than 1.0, explaining 81.1% of the variance in the data set. The results are shown in the table below.

After examining if there were any items that did not load strongly on any factor, that loaded on a factor other than the one intended, or that loaded relatively equally across multiple factors, an analysis of the loadings was conducted. Through Varimax rotation, the 18 items were cleanly loaded onto six factors – Perceived Ease of Use (PEU), Perceived Usefulness (PU), Social Influence (SI), Flow (F), Credibility (C) and Behavioral Intention to Use (BIU), as shown in the table (Rotated Component Matrix) below. All the items under BIU loaded heavily on Factor 1 (>=0.827), those under C loaded heavily on Factor 2 (>=0.833), those under PU loaded heavily on Factor 3 (>=0.702), those under SI loaded heavily on Factor 4 (>=0.689), those under PEU loaded heavily on Factor 5 (>=0.732), and those under F loaded heavily on Factor 6 (>=0.623).

Regression Analysis   

Linear regression analysis was used to test the hypotheses and allow further validation of the instrument. The table below shows the linear regression model for Behavioral Intention.

The test was conducted on the factors derived from Principal Component Analysis in order to establish the relationship of PEU, PU, SI, F and C with BIU. Factors 2 to 6 with their factor scores were used as independent variables in multiple regression analysis, and Factor 1 (BIU) was used as the dependent variable.

Standard Errors and t values of the regression coefficients for Factor 2, Factor 3, Factor 4, Factor 5 and Factor 6 are presented in table below. All the selected factors were found as significant (P<0.01). Regression coefficients of factors indicate that all factors had significant-positive linear relationships with Behavioral Intention to Use, i.e., Factor 1. 

The variance explained was very strong (R square=.977) with all the following coefficients found to be significant at p = .000: Perceived Ease of Use, Perceived Usefulness, Credibility, Social Influence and Flow. This provides strong statistical support H1, H2, H3, H4 and H5 respectively.

Also, Variance Inflation Factor (VIF) for each of the variables is low (<10), which indicates absence of multicollinearity.

#2.  Relative importance of purchase determinants

The following charts depict the relative importance of purchase determinants for different categories of Smartphone apps.

As shown in the plots, the most important determinants of purchase decision in each of the categories are:

Entertainment apps : Pleasure
Networking apps : Word of Mouth
Productivity apps : Usefulness
Infocs : Ease of Use and Usefulness



#3.  Clustering of Smartphone app users
 
Cluster Analysis

On running hierarchical clustering on users' response data, a two cluster solution was arrived at, with majority of cases choosing Productivity or Information apps as their most used type falling into Cluster 1; those choosing Networking or Entertainment falling into Cluster 2.

The final cluster centers are shown below:

The clusters may be interpreted as follows:

Cluster 1:   Those who use their Smartphone mostly for Productivity or Information apps. These users give high importance to Usefulness and Trial Performance of an app while purchasing it.

Cluster 2:   Those who use their Smartphone mostly for Entertainment or Networking apps. These users give high importance to Word of Mouth and Pleasure while purchasing an app.