Analysis of Mailer Impact on Purchase Intention

Nov 11, 2025, 4:18 PM

Kiblerly-Clarke Data csv.csv

Missing file

Step 1: Load and Clean the Data

Import the dataset into SPSS (or Python/JASP/Jamovi).
Identify and recode Don’t Know (DK) / blank responses as missing values.
Create a binary variable for Mailer_Opened:
1 = Opened
0 = Did Not Open
Exclude DK from analysis.
Ensure Likely_Purchase, Overall_Quality, Brand_I_Trust, and Brand_I_Recommend are correctly treated as numerical rating scales.
Step 2: Form Research Questions and Hypotheses
Use the mailer evaluation data to answer:
RQ # Research Question Variables Used Objective
RQ1 Does opening the mailer affect purchase intention? Mailer_Opened vs Likely_Purhcase Test message effectiveness
RQ2 Do quality and trust ratings drive recommendation (loyalty)? Overall_Quality, Brand_I_Trust → Brand_I_Recommend Identify key loyalty drivers
RQ3 Does child age influence purchase behavior? Age_In_Months vs Likely_Purhcase Segment by baby development stage
RQ4 How is the mailer perceived? Info_Appropriate, Info_Believable, Info_Understanding Evaluate communi

RQ 1 – Does opening the mailer affect purchase intention?

We compared the average Likely Purchase score between parents who opened the mailer and those who did not.
Mailer StatusNMean Likely PurchaseSD
Opened2821.721.05
Did Not Open183.001.14
Using an independent‐samples t-test (Welch correction):
t=−4.65        p=0.0002t = -4.65\;\;\;\; p = 0.0002
A significantly lower (better) purchase-likelihood score is observed among those who opened the mailer. In this 1–5 scale (1 = "Definitely will buy", 5 = "Definitely will not"), opening the mailer increases purchase intention.
The bar chart below visualises the difference including ±1 SD error bars.

RQ 1 – Does opening the mailer affect purchase intention?

We cleaned the data, created a binary Mailer_Opened flag, corrected the typo in Likely_Purhcase → Likely_Purchase, and converted the rating to numeric.
Findings
• Parents who opened the mailer (N = 282) had a mean Likely Purchase score of 1.72 ± 1.05.
• Those who did not open it (N = 18) averaged 3.00 ± 1.14.
(Scale: 1 = “Definitely will buy”, 5 = “Definitely will not buy”. Lower is better.)
A Welch two-sample t-test showed the difference is statistically significant:
t(≈21) = −4.65, p = 0.0002.
➡️ Opening the mailer significantly increased purchase intention.
A bar chart with ±1 SD error bars was displayed to illustrate the gap.
Would you like to:
  • explore the drivers of recommendation (RQ 2),
  • test the influence of child age on purchase (RQ 3), or
  • examine detailed perceptions of the mailer (RQ 4)?

tes

RQ 2 – Do quality and trust ratings drive recommendation (loyalty)?

We modelled Brand_I_Recommend (1 = Definitely would recommend … 5 = Definitely would not) as a function of
  • Overall_Quality
  • Brand_I_Trust
( lower scores = more favourable ).
Sample used: 300 complete cases.

1. Correlations

Overall_QualityBrand_I_TrustBrand_I_Recommend
Overall_Quality1.000.560.38
Brand_I_Trust0.561.000.66
Both predictors correlate with recommendation, with trust showing the strongest bivariate link ((r=0.66)).

2. Multiple regression results

Brand_I_Recommend=β0+β1 Overall_Quality+β2 Brand_I_Trust+ε\text{Brand\_I\_Recommend}=\beta_0+\beta_1\,\text{Overall\_Quality}+\beta_2\,\text{Brand\_I\_Trust}+\varepsilon
PredictorEstimate (β)tp
Intercept0.800.73.47
Overall_Quality0.030.26.80
Brand_I_Trust0.9912.6<.001
Model fit: R² = 0.44 (Adj. R² = 0.44), F(2, 297) = 117.1, p < .001.
➡️ Brand trust is the dominant driver of recommendation. When trust is accounted for, perceived quality adds no unique explanatory power.

3. Visual check

A scatter plot of Actual vs. Predicted recommendation (shown above) confirms the model’s reasonable fit and illustrates the linear trend.
Would you like to:
  • analyse how child age influences purchase intention (RQ 3),
  • dig deeper into mailer perception items (RQ 4), or
  • explore any other question?

RQ 2 – Do quality and trust ratings drive recommendation?

We modelled Brand_I_Recommend (1 = “Definitely would recommend” … 5 = “Definitely would not”) as a function of Overall_Quality and Brand_I_Trust using 300 complete cases.
Key findings
• Correlation: Brand_I_Trust correlates most strongly with recommendation (r = 0.66).
• Regression:
– Overall_Quality → β = 0.03, p = .80 (not significant)
– Brand_I_Trust → β = 0.99, p < .001 (highly significant)
Model R² = 0.44.
➡️ Brand trust is the primary loyalty driver; perceived quality adds no unique impact once trust is considered.
A plot of predicted vs. actual recommendation scores confirms the linear fit.
Feel free to ask next:
  • “Explore the influence of child age on purchase intention (RQ 3).”
  • “Show detailed perceptions of the mailer (RQ 4).”
  • “Provide an overall executive summary of the findings so far.”
Get started with Vizly