Skip to content
Jisan Nuren Siam
Jisan Nuren Siam

Data · Research · Products

All work

Data Science & AnalyticsSPSS · Regression analysis

Online Shopping Behavior Analysis

Which psychological factors actually move online buying behavior? A survey study of 196 online shoppers in Pakistan, analyzed in SPSS.

Overview graphic: 196 respondents, 30 measurement items, a 7-point Likert scale, six constructs and five hypotheses, with a conceptual model linking five factors to consumer buying behavior.
Fig. 01 — Project overview
Role
Statistical analysis in SPSS: reliability and validity checks, diagnostics, multiple regression and hypothesis decisions.
Type
Professional work
Status
Completed analysis
Tools
SPSS 26 · Multiple regression · Reliability analysis
Respondents
196
Constructs
6 · 30 items
Scale
7-point Likert
Software
SPSS 26

Case study

01Question

Five factors, one outcome

The study tested trust, sales promotion, information quality, perceived risk and social influence as predictors of consumer buying behavior.

02Method

From survey to model

  • Data preparation and assumption checks
  • Reliability (Cronbach's alpha) and convergent validity (AVE)
  • Descriptive and correlation analysis
  • Multiple linear regression with VIF, Cook's distance and Durbin–Watson diagnostics
Seven-step analysis pipeline from survey design and data preparation through reliability review, descriptive analysis and regression diagnostics to hypothesis decisions.
Fig. 02 — Analysis pipeline

View full size: Analysis pipeline (opens in a new tab)

03Measurement

Reliable constructs

Cronbach's alpha ranged from .806 to .906 and AVE from .568 to .732 across the six constructs.

Table of Cronbach's alpha and AVE for the six constructs, noting all alpha values exceed .80 and all AVE values exceed .50.
Fig. 03 — Measurement quality

View full size: Measurement quality (opens in a new tab)

04Findings

Information quality led

The model explained 66.9% of the variance in buying behavior (R² = .669, adjusted R² = .660; F(5,190) = 76.794, p < .001). Information quality (β = .410), social influence (β = .307) and trust (β = .245) were positive predictors, and perceived risk was negative (β = −.129, p = .009). Sales promotion was not significant at conventional levels (β = .095, p = .098).

Bar chart of standardized regression coefficients for the five predictors, with information quality strongest, perceived risk negative and sales promotion not significant; R² = .669 and four of five hypotheses supported.
Fig. 04 — Regression results

View full size: Regression results (opens in a new tab)

05Validation

Diagnostics before conclusions

Maximum VIF (2.654), maximum Cook's distance (.595) and Durbin–Watson (1.854) were reviewed before the results were interpreted. The model describes associations in this sample; it does not establish cause.

Model fit and diagnostics panel with R², adjusted R², F, p-value and Durbin–Watson, maximum VIF and Cook's distance, and a hypothesis summary table.
Fig. 05 — Model validation

View full size: Model validation (opens in a new tab)