Review Reports - Opening Mechanisms and Visual Configurations in Cosmetic E-Commerce Packaging: Effects on Purchase Intention

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The study addresses a timely and practically relevant topic; several methodological and reporting issues limit the robustness and generalizability of the findings. Please find my comments below:

The title is lengthy and lists several design attributes in parallel. Consider shortening it to highlight the study’s central contribution, such as how e-commerce packaging design influences purchase intention.
The abstract refers to a “two-part experiment,” whereas the methods section describes a single within-subjects experiment with six packaging conditions. Also, correct the finding on printing: the results indicate that offset printing, not watermark printing, increased purchase intention relative to watermark printing.
Make the link between the growth of e-commerce and the specific research gap more explicit. Explain what remains unknown about how e-commerce packaging shapes consumer evaluations or purchase intentions.
Explain more clearly why opening mechanisms, brand identity cues, and printing approaches were selected, describing their practical relevance and theoretical links to consumer perceptions. The literature review would also benefit from less citation clustering and more synthesis of what prior studies establish and how this study advances that work.
H1b combines perceived quality and brand image in one prediction. Separate these into distinct hypotheses to make the predictions and tests clearer. Also explain how perceived quality from perceived brand image and why both are needed as mediators.
The reported power analysis appears to assume a paired-samples t-test, while the study includes comparisons across six conditions and multiple mediators. Clarify the basis and limits of the sample size calculation, including its implications for power and multiple testing.
Report the specific scale items and reliability and validity evidence for the adapted measures. Because purchase intention is measured with a single item, provide a rationale for that choice. Include the measures in an appendix.
Printing approach was manipulated only when brand identity was present. As a result, the effects of printing approach and brand identity cannot be fully separated. Acknowledge this limitation and explain how it affects interpretation; a future study could use a fully orthogonal design.
Describe how presentation order was randomized and whether order effects were assessed. Report the findings of any order-effect analysis.
The study is described as within-subjects, but the opening-mechanism comparison is reported using an independent-samples t-test. Reanalyse it using an appropriate paired-samples test. Also, verify the reported degrees of freedom: with 41 participants in a paired comparison, the expected degrees of freedom are 40, not 80. Correct the associated typographical error in the p-value.
The three-level ANOVA appears to combine brand identity and printing approach, even though they are distinct manipulations and printing was conditional on brand identity being present. Separate and describe the analyses accordingly, or clearly acknowledge the design constraint and limit the conclusions to the comparisons the design supports.
Explain how the analysis addresses the number of statistical tests and the risk of inflated Type I error; apply an appropriate correction or justify the approach. The mediation analyses are reported separately for each mediator. Consider testing the mediators together in a parallel mediation model, or explain why separate models are appropriate and what they do not establish.
Integrate the word-association analysis more clearly with the research questions and hypothesis findings. Explain what it adds to the interpretation of the experimental results.
Lead with the main findings and contribution rather than restating the study’s purpose. Reframe the claim that the study “introduces” perceptual fluency theory: the contribution is its application to e-commerce packaging. Engage more directly with any findings that differ from prior research and make the practical recommendations more specific to the results.
Expand the discussion of the simulated packaging stimuli. Clean, lab-prepared stimuli may overstate the influence of visual cues and omit the effects of shipping damage, dirt, or wear, limiting the ecological validity of the findings.
Use one term consistently for the packaging construct, and edit the manuscript for phrasing, grammar, and typographical errors. In Table 2, use consistent notation and explain when printing approach is not applicable. In Figure 7, identify the axes and consolidate the interpretation in the text or caption. Apply one reference style consistently.

Author Response

Dear Editors and Reviewers,

Thank you for your constructive comments on our manuscript entitled “The Impact of Opening Mechanisms, Printing Approaches, and Brand Identity Cues in Cosmetic E-commerce Packaging on Purchase Intention” (Manuscript ID: jtaer-4638302). Your feedback has helped us strengthen the manuscript.

We have carefully considered each comment and revised the manuscript accordingly. For ease of review, revisions addressing comments specific to Reviewer 1 are shown in red text, while those addressing comments specific to Reviewer 2 are shown in blue text. Revisions addressing overlapping comments from both reviewers are shown in red text.

Below, we provide our point-by-point responses to each reviewer’s comments.

Reviewer #1:

Q1 The title is lengthy and lists several design attributes in parallel. Consider shortening it to highlight the study’s central contribution, such as how e-commerce packaging design influences purchase intention.

Response：Thank you for this suggestion. We have shortened the title to “Opening Mechanisms and Visual Configurations in Cosmetic E-commerce Packaging: Effects on Purchase Intention.” The revised title highlights the two experimental factors and the focal outcome while retaining the cosmetic e-commerce context.

Location: Title page, p. 38.

Q2 The abstract refers to a “two-part experiment,” whereas the methods section describes a single within-subjects experiment with six packaging conditions. Also, correct the finding on printing: the results indicate that offset printing, not watermark printing, increased purchase intention relative to watermark printing.

Response：Thank you for identifying these inconsistencies. We have revised the abstract to describe a single 2 × 3 fully within-subjects experiment in which all 41 participants opened and evaluated all six packaging conditions. Word associations and questionnaire ratings were collected within this experiment, rather than in two separate experiments. We have also corrected the visual comparison: purchase intention was higher for V3, featuring offset printing with a white logo on a black background, than for V2, featuring watermark printing with a black logo on natural cardboard. The abstract now clarifies that this finding concerns the tested visual configurations, because printing approach, brand logo presence, and color were not independently manipulated.

Location: Abstract, pp. 38–39.

Q3 Make the link between the growth of e-commerce and the specific research gap more explicit. Explain what remains unknown about how e-commerce packaging shapes consumer evaluations or purchase intentions.

Response：Thank you for this suggestion. We have revised the final paragraphs of the Introduction to make the research gap more explicit. Building on the discussion of e-commerce growth and consumers’ physical encounters with packaging during delivery and unboxing, we now clarify what remains less understood: how the opening experience and visual presentation of cosmetic e-commerce packaging relate to evaluations of the enclosed product, impressions of the brand, and purchase intention within the same experimental setting. We then explain how the study addresses this gap by examining two opening mechanisms and three visual configurations, together with the proposed indirect associations through perceived opening fluency, perceived product quality, and perceived brand image.

Location: Introduction, final two paragraphs, pp. 43–44.

Q4 Explain more clearly why opening mechanisms, brand identity cues, and printing approaches were selected, describing their practical relevance and theoretical links to consumer perceptions. The literature review would also benefit from less citation clustering and more synthesis of what prior studies establish and how this study advances that work.

Response：Thank you for this suggestion. We have revised the theoretical discussion to clarify the practical relevance and theoretical rationale for examining these packaging attributes. Opening mechanisms concern consumers’ physical access to the enclosed product and the ease of the opening experience. Brand logos provide information for brand recognition and evaluation, while printing approaches contribute to the reproduction of packaging colors and graphics. These attributes therefore connect the physical interaction and visual communication aspects of cosmetic e-commerce packaging.

We have also strengthened the synthesis of prior research. Section 2.4 distinguishes technical findings about printed output from consumer research on packaging cues, product judgments, and brand impressions, and explains how these complementary bodies of evidence motivate the V3–V2 configuration comparison. Section 2.5 distinguishes general processing fluency from perceived opening fluency and develops the rationale for the proposed indirect associations through product quality and brand image evaluations. Throughout these revisions, we clarify that the visual comparisons concern combinations of logo information, printing approach, and color, rather than independently manipulated attributes. These changes connect the literature more explicitly to the study’s constructs and hypotheses without treating technical printing evidence as direct evidence of consumer responses.

Location: Sections 2.1–2.5, pp. 44–54, particularly Sections 2.4–2.5, pp. 49–54.

Q5 H1b combines perceived quality and brand image in one prediction. Separate these into distinct hypotheses to make the predictions and tests clearer. Also explain how perceived quality from perceived brand image and why both are needed as mediators.

Response：Thank you for this suggestion. We have separated the original combined prediction into H1b, concerning perceived product quality, and H1c, concerning perceived brand image. The opening-mechanism prediction for purchase intention is now numbered H1d.

We have also clarified the conceptual distinction in Section 2.5. Perceived product quality concerns consumers’ evaluations of the enclosed product inferred from its packaging, whereas perceived brand image concerns evaluations of the brand, operationalized through perceived industry leadership, experience, and customer orientation. Both are included as proposed mediators because packaging may inform judgments about the product and the organization behind it. Treating them separately allows us to examine these distinct proposed indirect associations with purchase intention. The parallel mediation models estimate each specific indirect effect while accounting for the other mediators included in the model. Figure 1 has been updated to reflect the revised hypothesis numbering.

Location: Opening-mechanism hypotheses in Section 2.2, p. 47; Section 2.5, pp. 51–54; Figure 1, p. 55.

Q6 The reported power analysis appears to assume a paired-samples t-test, while the study includes comparisons across six conditions and multiple mediators. Clarify the basis and limits of the sample size calculation, including its implications for power and multiple testing.

Response：Thank you for this comment. We have revised the sample-size justification to distinguish the requirements for the two main effects from those of the broader analyses. Using the “ANOVA: Repeated measures, within factors” procedure in G*Power 3.1, sample-size requirements were evaluated separately for each main effect using participant-level scores averaged across the other factor. Assuming f = 0.25, α = 0.05, target power = 0.80, a repeated-measures correlation of 0.50, and ε = 1, the required sample sizes were 34 participants for opening mechanism and 28 for visual configuration. The final sample of 41 participants exceeded the larger estimate under these assumptions.

These calculations concern the two main effects and do not establish adequate power for interaction effects, indirect effects, or all multiplicity-adjusted tests. We have therefore limited the sample-size claim accordingly. Multiple-testing procedures are described separately: Holm adjustment covers the 12 ANOVA omnibus tests, Bonferroni adjustment covers the three visual pairwise comparisons within each outcome, and separate Bonferroni adjustments cover the nine indirect effects in each mediation specification. Effect sizes and confidence intervals are reported where applicable to communicate the magnitude and precision of the estimates.

Location: Section 4.1, pp. 63–64; Section 5.2 and Tables 5–8, pp. 80–90.

Q7 Report the specific scale items and reliability and validity evidence for the adapted measures. Because purchase intention is measured with a single item, provide a rationale for that choice. Include the measures in an appendix.

Response： We have expanded Section 4.2 to describe the constructs, item counts, scoring procedures, and response anchors. Appendix A, Table A1 presents English translations of all administered measurement items and their anchors. The item sources and adaptations are described in Section 4.2.

Internal consistency was assessed separately for each packaging condition. Cronbach’s α ranged from 0.860 to 0.954 for the five-item perceived product quality measure and from 0.841 to 0.962 for the three-item perceived brand image measure; all analyses included 41 participants. Condition-specific coefficients are reported in Appendix A, Table A2. These results support internal consistency but do not establish construct validity or measurement invariance.

The single purchase-intention item directly targets a specific judgment: participants’ stated willingness to purchase products from the brand after experiencing its packaging. It provides a focused assessment of this intention rather than a broader measure of purchasing behavior. We acknowledge that a single item limits construct coverage and does not permit assessment of internal consistency. Section 6.3 now identifies these limitations for both purchase intention and perceived opening fluency and recommends further measurement development and validation.

Location: Section 4.2, pp. 64–65; Appendix A, Tables A1–A2, pp. 107–108; Section 6.3, p. 97.

Q8 Printing approach was manipulated only when brand identity was present. As a result, the effects of printing approach and brand identity cannot be fully separated. Acknowledge this limitation and explain how it affects interpretation; a future study could use a fully orthogonal design.

Response：Thank you for highlighting this limitation. We agree that brand logo presence and printing approach were not independently manipulated. We have therefore described the visual factor as three configurations: V1 comprised unprinted natural cardboard without a brand logo; V2 displayed a black logo on natural cardboard using watermark printing; and V3 displayed a white logo on a black background using offset printing.

The V2–V1 comparison combines the introduction of a brand logo with printing and cannot identify an independent effect of logo presence. The V3–V2 comparison involves changes in printing approach, background color, and logo color and cannot isolate an effect of printing technology. We have revised the hypotheses and interpretation of the findings to refer to these complete configurations rather than separate brand-identity or printing effects. Section 6.3 explicitly acknowledges this limitation and recommends future experiments that independently manipulate these attributes in an orthogonal design.

Location: Section 3.2 and Table 2, pp. 58–62; Sections 2.3–2.4, pp. 47–50; Section 5.2.2, pp. 84–86; Section 6.3, p. 97; Section 7, pp. 99–100.

Q9 Describe how presentation order was randomized and whether order effects were assessed. Report the findings of any order-effect analysis.

Response：Thank you for this comment. We have clarified in Section 4.3 that each participant was assigned a computer-generated random sequence for opening and evaluating the six packaging conditions. Participant-level presentation orders were recorded and used to assess presentation-position effects.

For each outcome, we fitted linear mixed models with packaging condition and presentation position (1–6) as categorical fixed effects and a participant-specific random intercept. Models were estimated using maximum likelihood, and likelihood-ratio tests compared models with and without presentation position. The four position-effect tests formed a separate Holm-adjusted family.

After accounting for packaging condition, no significant overall presentation-position effects were detected for purchase intention, χ² (5) = 6.196, p = 0.288; perceived product quality, χ² (5) = 1.038, p = 0.959; perceived brand image, χ² (5) = 1.528, p = 0.910; or perceived opening fluency, χ² (5) = 3.179, p = 0.672. All Holm-adjusted p values were 1.000. These findings are reported in Section 5.2 and Appendix A, Table A3. We interpret them as a lack of detected overall position effects, rather than proof that order effects are absent; condition-specific carryover effects were not assessed.

Location: Section 4.3, p. 66; Section 5.2, p. 79; Appendix A, Table A3, p. 108.

Q10 The study is described as within-subjects, but the opening-mechanism comparison is reported using an independent-samples t-test. Reanalyse it using an appropriate paired-samples test. Also, verify the reported degrees of freedom: with 41 participants in a paired comparison, the expected degrees of freedom are 40, not 80. Correct the associated typographical error in the p-value.

Response：Thank you for identifying this error. We have replaced the independent-samples comparison with 2 × 3 repeated-measures ANOVAs for the four outcomes, treating opening mechanism and visual configuration as within-subject factors. All 41 participants contributed observations to all six conditions. For the two-level opening-mechanism factor, the ANOVA main-effect test corresponds to the paired comparison of participant-level scores averaged across the three visual configurations.

The opening-mechanism tests now have degrees of freedom of 1 and 40. For purchase intention, the corrected result is F (1, 40) = 32.002, Holm-adjusted p < 0.001, partial η² = 0.444. The estimated marginal mean difference for hand-opening minus tool-assisted opening is 0.894, with a 95% confidence interval of [0.575, 1.214]. We have corrected the associated p-value reporting. Tables 4–6 now present descriptive statistics, repeated-measures ANOVA results, and estimated marginal mean comparisons, with the applicable multiplicity adjustments specified.

Location: Section 5.2, pp. 77–79, particularly Section 5.2.1, pp. 79–84; Tables 4–6 and their notes, pp. 78–84.

Q11 The three-level ANOVA appears to combine brand identity and printing approach, even though they are distinct manipulations and printing was conditional on brand identity being present. Separate and describe the analyses accordingly, or clearly acknowledge the design constraint and limit the conclusions to the comparisons the design supports.

Response：We agree that the original factor label did not adequately distinguish the tested configurations from independently manipulated attributes. We have addressed this by explicitly acknowledging the design constraint and restricting the conclusions to comparisons supported by the experiment.

The revised analysis uses a three-level visual-configuration factor crossed with the two-level opening-mechanism factor in a fully within-subjects design. V1–V3 represent the actual combinations of brand logo presence, printing approach, and color described in Section 3.2 and Table 2. The visual-configuration main effect therefore tests differences among these combinations, rather than independent effects of brand identity or printing. Estimated marginal mean comparisons report all three pairwise contrasts—V2–V1, V3–V2, and V3–V1—averaged across opening mechanisms, with Bonferroni adjustment within each outcome. The hypotheses, results, discussion, and conclusions now interpret these comparisons at the configuration level, and Section 6.3 acknowledges the need for future independent manipulation of the constituent attributes.

Location: Section 3.2 and Table 2, pp. 58–62; Sections 5.2–5.2.2 and Tables 5–6, pp. 77–86; Section 6.3, p. 97; Section 7, pp. 99–100.

Q12 Explain how the analysis addresses the number of statistical tests and the risk of inflated Type I error; apply an appropriate correction or justify the approach. The mediation analyses are reported separately for each mediator. Consider testing the mediators together in a parallel mediation model, or explain why separate models are appropriate and what they do not establish.

Response We have clarified the testing families and implemented multiplicity adjustments. Holm’s procedure covers the 12 ANOVA omnibus tests, comprising two main effects and their interaction for each of four outcomes. The three visual pairwise comparisons form a separate Bonferroni-adjusted family within each outcome. Each outcome has one opening-mechanism comparison, so its Bonferroni adjustment leaves the p value and confidence interval unchanged.

We have retained the single-mediator analyses and added within-subject parallel mediation models to examine the proposed mediators simultaneously. The opening-mechanism parallel model includes perceived opening fluency, perceived product quality, and perceived brand image. Each of the three visual-configuration models—V2–V1, V3–V2, and V3–V1—includes perceived product quality and perceived brand image simultaneously. Outcome regressions include mediator difference scores and mean-centered mediator-average terms.

Percentile bootstrap inference uses 50,000 participant-level resamples with seed 20261016. The nine specific indirect effects from the parallel models constitute one Bonferroni-adjusted family; the nine single-mediator indirect effects constitute another. Each uses a confidence level of 99.444444% (approximately 99.4%), corresponding to a familywise α of 0.05. These separate adjustments do not provide joint familywise error control at 0.05 across both families. Total indirect effects, direct effects, total effects, and individual regression paths are outside these adjustment families.

The revised interpretation distinguishes associations estimated in single-mediator models from specific indirect effects estimated while accounting for other mediators in parallel models. Differences in whether intervals exclude zero are not interpreted as statistically significant differences between model estimates. In the parallel models, the opening-mechanism indirect association through perceived product quality met the adjusted criterion; the remaining specific indirect effects did not.

Location: Section 5.2, p. 77; Section 5.2.3, pp. 86–94; Tables 5–8 and their notes, pp. 80–90.

Q13 Integrate the word-association analysis more clearly with the research questions and hypothesis findings. Explain what it adds to the interpretation of the experimental results.

Response：Thank you for this suggestion. We have revised Sections 5.1 and 6 to connect the word-association findings more explicitly with the research questions and questionnaire results. Collected before the structured ratings, the word associations identify aspects of packaging that featured spontaneously in participants’ impressions.

The Discussion now explains two connections. First, references to opening ease and difficulty show that the opening experience was salient to participants, providing context for the higher perceived opening fluency ratings in the hand-opening condition. Second, the prominence of appearance-related associations and the greater frequency of quality-related associations for V3 than V1 help contextualize the differences in product quality, brand image, and purchase intention ratings across visual configurations. These qualitative findings add information about the content and salience of participants’ impressions beyond the questionnaire scores. We interpret them as exploratory context for the experimental findings, rather than independent tests of the hypotheses or evidence establishing the proposed mediation mechanisms.

Location: Section 5.1, pp. 67–77; Figures 5–6, pp. 71–73; Discussion, pp. 91–94.

Q14 Lead with the main findings and contribution rather than restating the study’s purpose. Reframe the claim that the study “introduces” perceptual fluency theory: the contribution is its application to e-commerce packaging. Engage more directly with any findings that differ from prior research and make the practical recommendations more specific to the results.

Response：Thank you for this suggestion. We have revised the opening of the Discussion to move from a brief statement of scope to the main findings: hand-opening received more favorable ratings across all four outcomes, while perceived product quality, perceived brand image, and purchase intention followed the ordering V3 > V2 > V1. The contribution is framed as a contextual application of the S-O-R framework to cosmetic e-commerce packaging, rather than the introduction of perceptual fluency theory or a general theoretical extension.

The revised Discussion also distinguishes the higher perceived opening fluency observed for hand-opening from the lack of multiplicity-adjusted support for its indirect association with purchase intention in both mediation specifications. We relate this finding to prior fluency research while explaining that the ease of physically opening packaging differs from the information-processing ease examined in that literature.

Practical recommendations are now tied to the observed results, focusing on hand-opening designs and the tested branded, printed visual configurations. We explicitly avoid attributing the visual differences to individual attributes. Unsupported claims about cost reduction and improved sustainability have been removed, and potential implications for shopping satisfaction, brand trust, and longer-term emotional connections are explicitly identified as requiring further investigation.

Location: Discussion and Sections 6.1–6.2, pp. 94–97.

Q15 Expand the discussion of the simulated packaging stimuli. Clean, lab-prepared stimuli may overstate the influence of visual cues and omit the effects of shipping damage, dirt, or wear, limiting the ecological validity of the findings.

Response：Thank you for highlighting this ecological-validity limitation. We have expanded Section 6.3 to clarify that participants physically opened and evaluated laboratory-prepared packaging that had not undergone actual logistics transportation. Consequently, the cleanliness and physical condition of the stimuli may differ from those of packages received through routine delivery. We also qualify the interpretation of “clean” in the word-association results, noting that its occurrence may partly reflect the preparation of the experimental materials. These limitations restrict generalization to everyday delivery experiences. We recommend examining packaging after actual shipment to assess whether the observed differences persist under realistic delivery and unboxing conditions, and reiterate this direction in the Conclusions.

Location: Section 5.1.1, discussion of “clean,”, pp. 68–69; Section 6.3, first limitation, p. 97; Section 7, p. 99.

Q16 Use one term consistently for the packaging construct, and edit the manuscript for phrasing, grammar, and typographical errors. In Table 2, use consistent notation and explain when printing approach is not applicable. In Figure 7, identify the axes and consolidate the interpretation in the text or caption. Apply one reference style consistently.

Response：We have standardized the key terms used for the experimental factors and measured constructs: opening mechanism, visual configuration, perceived opening fluency, perceived product quality, perceived brand image, and purchase intention. We have also revised phrasing, grammatical errors, and typographical inconsistencies in the manuscript.

Table 2 now uses “N/A” for the printing approach in the unprinted V1 conditions. Its note explains that “N/A” means “not applicable because the packaging was unprinted,” distinguishing the absence of printing from a printing technology. We have also standardized the numerical reference format and removed duplicate reference entries, including the duplicated Pousette, Hensler, Dodds, and Reber records.

Figure 7 identifies the horizontal and vertical axes as Dimension 1 (49.40%) and Dimension 2 (29.88%), respectively. We have clarified the caption to identify the nine semantic dimensions and the six packaging conditions and to report the inertia explained by each axis. The interpretation is presented in Section 5.1.2 rather than repeated in the caption.

Location: Throughout the manuscript, pp. 38–107; Table 2 and its note, p. 62; Section 5.1.2 and Figure 7, pp. 73–77; References, pp. 101–106.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript, which is called "The Impact of Opening Mechanisms, Printing Approaches, and Brand Identity Cues in Cosmetic E-commerce Packaging on Purchase Intention," focuses on an important topic that lies at the crossroads of e-commerce, packaging, and consumer behavior. The incorporation of the physical packaging stimulus into the S-O-R model seems to have some promising prospects, and the use of the physical packaging stimuli together with the word association method can be considered a positive point of this research.

At the same time, the manuscript contains a number of serious drawbacks regarding data collection and analysis. These problems go beyond mere editing and translation and require thorough reconsideration of the research design.

Major Comments

Inadequate statistical analysis of the within-subjects experiment

The most serious issue concerns the contradiction between the experimental design and the methods of statistical analysis. Section 4.1 provides information about a within-subjects experiment, where 41 participants were presented with six packaging configurations.

However, Section 5.2 shows independent-samples t-tests for opening mechanisms t(80)) and one-way ANOVA for brand identity conditions F(2,120)). These tests imply treating observations made from the same participants as independent ones. Also, Section 5.2.3 describes the usage of paired-samples t-tests with the degrees of freedom of 80 and 40 being reported.

This raises substantial doubts regarding the assumption of independence, estimates of standard errors, and reliability of the significance levels reported.

The authors should reanalyze their data with regard to the proper analysis for a within-subjects experiment, i.e., repeated-measures ANOVA or linear mixed-effects models with participant-level random effects. It should take into account the six experimental conditions and dependencies between observations. The contrasts, effect sizes, confidence intervals, and multiple comparison adjustment should be also reported.

Confounding of printing approaches and packaging color

The printing condition does not allow to isolate printing technology as a factor having causal impact. According to Section 3.2, the watermark printing was exemplified by a natural corrugated cardboard with black lettering, whereas the offset printing was exemplified by a black package with white lettering.

As a result, experimental conditions differ not only in printing technology, but also in color, contrast, and appearance of the package.

The reported differences thus cannot be attributed exclusively to printing technology. They can be caused by preferences toward color, visual contrast, premium position, or interaction between these variables.

Although this problem is mentioned in Section 6.3, it is insufficiently considered.

A better-designed experiment should use printing technology as a factor while keeping other visual variables constant. Alternatively, the findings of this paper should be interpreted as comparing complete visual packages, rather than isolated printing effects.

Incomplete separation of the effects of brand identity and printing

The six experimental conditions represent incomplete factorial design, 2 × 2 × 2. The absence of brand identity implies also the absence of printing, whereas both printing conditions include brand identity.

Although this decision can be dictated by practical packaging constraints, it creates certain difficulties for estimation of the effects separately attributed to brand identity and to printing.

The three-level classification of the factors in Section 5.2.2—no logo, watermark printing, and offset printing—is based on combination of the two different factors rather than on their separate manipulation.

The authors should make clear which causal contrasts can be established with the help of the current design. They should limit their claims about independent effects and redesign factorial experiment in order to provide stronger evidence for the theoretical links suggested in the manuscript.

Insufficient reliability of the mediation analysis

The mediation analyses presented in Tables 4-10 require substantial reconsideration.

First of all, it is unclear whether the authors took into account the repeated-measures structure of their experiment in their analysis. Standard regression-based procedures for conducting mediation analysis can provide misleading estimates of uncertainty in case if this structure is ignored.

Secondly, the proposed mediators and purchase intention seem to be measured at the same evaluation episode, which complicates causal mediation claim.

There are also some numerical discrepancies in the reported results:

Table 4 shows an indirect effect of 0.164, whereas the corresponding text mentions 0.172.

In the analysis of the effect of opening mechanism on perceived quality, text reports a negative indirect effect of approximately −0.635, whereas Table 5 shows the positive effect of 0.635. Moreover, the discussion mentions a negative direct effect, although the positive coefficient is reported in Table 5.

The text before Table 8 states path a equal to 0.049, whereas Table 8 reports 1.049.

Some of the indirect effects are not consistent with the product of path coefficients shown in Tables 7 and 10.

This raises doubts about model specification and interpretations and reporting of those interpretations.

The authors need to provide the full equations for mediation, their coding scheme, method of estimation and calculation of confidence intervals. They must also deal with the situation where there is more than one mediator at the same time.

Sample size, statistical power, and generalizability

The study uses 41 participants recruited via convenience sampling in Ningbo, China. The sample size is relatively small and seems to be concentrated among university students, including those who have design education.

Although the power analysis has been conducted, it assumes the medium effect size for the paired comparison, which does not necessarily ensure statistical power for the whole set of contrasts, mediators, and relationships considered.

The authors should provide a more specific power analysis, justify the precision of the estimates.

Also, limited demographic and educational diversity limits the generalizability of the findings to other cosmetics e-commerce population.

Measurement validity and construct operationalization

Section 4.2 provides insufficient information about measurement items, scale adaptation, and psychometric properties.

Purchase intention is measured with a single item, whereas the other constructs are measured with adapted scales. The manuscript does not sufficiently validate the measurement instruments used.

Also, there is a need for the justification of the difference between perceptual processing fluency and physical ease of opening.

The authors should report the complete measurement items, construct operationalization, reliability statistics, and validation.

Additionally, since the participants evaluate packages after opening them physically, it is needed to specify which kind of purchase intention is measured, i.e., the initial, future, or repurchase intention. These concepts are not the same.

Inconsistency in the qualitative word association analysis

Section 5.1 mentions 738 initial associations and 732 valid associations, whereas the number of respondents mentioned is 40 despite the fact that the experiment involves 41 participants.

The manuscript should clarify whether there is any participant who was excluded from the qualitative analysis and what are the criteria for this exclusion.

Besides, the frequencies shown in Figure 6 sum up to 566, whereas 732 valid associations are stated in the beginning. Percentage of 34.5% for visual perception seems to be calculated on the basis of total classified associations, rather than on the basis of 732 valid associations. Treatment of other associations should be clarified.

Although the associations were independently coded by three researchers, the manuscript does not mention the measure of intercoder agreement before consensus discussion.

The authors should provide a detailed coding protocol, clarify the relationship between the initial and final numbers of associations and report appropriate measure of intercoder agreement.

Contradictions between hypotheses, results, and interpretation

Several contradictions in the manuscript need correction.

In particular, the abstract states the positive impact of watermark printing on purchase intention, while Section 5.2.3 finds higher purchase intention in the case of offset printing.

Also, the printing hypothesis concerning purchase intention is designated as H3b in the theoretical section and H3c in the results section.

The H3a hypothesis described in Section 5.2.3 is formulated wrongly as a hypothesis concerning brand identity, not printing.

These contradictions should be removed through a careful audit of hypotheses and results.

Exaggeration of theoretical and practical implications

The manuscript claims contributions to the S-O-R framework and perceptual fluency theory. Right now, the theoretical contribution seems to be the application of the existing theories to a certain context of packaging, which does not seem to constitute any theoretical extension.

The distinction between the application and the theoretical advance should be clearer.

Several practical implications seem to be exaggerated. In particular, the reduced cost of packaging, improved sustainability, loyalty of customers and long-term benefits for brands are not confirmed by the performed experiment.

These implications should be deleted or at least stated as assumptions.

Figures and tables

Figures 5–7 can be made more clear in terms of the interpretation of the correspondence analysis. The tables 4-10 should have consistent terminology, correct coefficients, correct significance levels, and complete model description.

Accuracy and consistency of references

There are duplicates in the reference list separated by unnecessary year suffixes: Dodds et al. (1991a/1991b), Hensler et al. (2015a/2015b), Pousette et al. (2014a/2014b), and Reber et al. (2004a/2004b).

These duplicates should be removed and there should be a review of the citations in relation to their accuracy and relevance. The references related to the technology of printing seem to be primarily about graphics and perception of packaging by consumers.

English language and editorial consistency

The manuscript needs language editing. The repetitions in text, inconsistencies, grammar mistakes and inappropriate phrasing are evident in the manuscript. For instance, results concerning printing approaches are stated twice in section 5.2.3.

The terminology should be consistent in the manuscript in terms of opening approaches, printing approaches, perceived fluency and purchase intention.

Author Response

Reviewer #2:

Q1 Inadequate statistical analysis of the within-subjects experiment

This raises substantial doubts regarding the assumption of independence, estimates of standard errors, and reliability of the significance levels reported.

Response：Thank you for identifying this fundamental issue. We agree that the original analyses did not adequately account for the repeated-measures design. We have replaced them with separate 2 × 3 repeated-measures ANOVAs for the four outcomes, treating opening mechanism and visual configuration as within-subject factors. All 41 participants evaluated all six conditions, and participants—not individual condition ratings—are treated as the independent units of analysis.

The models test both main effects and their interaction. Opening-mechanism effects have degrees of freedom of 1 and 40; visual-configuration and interaction effects have uncorrected degrees of freedom of 2 and 80. Sphericity is assessed where applicable, with Greenhouse–Geisser corrections applied when violated, including the visual-configuration effect on perceived opening fluency.

Tables 4–6 now report descriptive statistics, omnibus tests with partial eta squared, and estimated marginal mean differences with confidence intervals. Opening mechanisms are compared across visual configurations, and visual configurations are compared across opening mechanisms. Table 6 explicitly identifies the contrast directions. Holm adjustment covers the 12 omnibus tests, and the three visual pairwise comparisons are Bonferroni-adjusted separately within each outcome. This reanalysis replaces the previous inference that incorrectly treated repeated observations as independent and corrects the associated degrees of freedom and p-value reporting.

Location: Section 5.2, pp. 77–90; Tables 4–6 and their notes, pp. 78–84.

Q2 Confounding of printing approaches and packaging color

As a result, experimental conditions differ not only in printing technology, but also in color, contrast, and appearance of the package.

Although this problem is mentioned in Section 6.3, it is insufficiently considered.

Response：Thank you for highlighting this confounding. We agree that the V3–V2 comparison cannot isolate the effect of printing technology: V3 combines offset printing with a white logo on a black background, whereas V2 combines watermark printing with a black logo on natural cardboard. Printing approach, background color, logo color, and visual contrast therefore vary together.

We have revised the manuscript beyond the limitations section to interpret this comparison consistently at the configuration level. Section 2.4 and H3a–H3c now concern V3 versus V2, and the Results, Discussion, and Conclusions attribute the observed differences to the configurations as presented rather than to offset printing alone. We do not identify color preferences, contrast, or perceived premium positioning as demonstrated explanations for these differences. Section 6.3 recommends future experiments that vary printing technology while holding other visual attributes constant, or independently manipulate these attributes to distinguish their contributions.

Location: Section 2.4 and H3a–H3c, pp. 49–50; Section 3.2 and Table 2, pp. 58–62; Section 5.2.2, pp. 84–86; Discussion, pp. 91–94; Section 6.3, p. 97; Section 7, pp. 99–100.

Q3 Incomplete separation of the effects of brand identity and printing

Although this decision can be dictated by practical packaging constraints, it creates certain difficulties for estimation of the effects separately attributed to brand identity and to printing.

Response：We agree that the study is not a fully crossed experiment independently manipulating opening mechanism, brand identity, and printing approach. We now describe the implemented design as a 2 (opening mechanism) × 3 (visual configuration) fully within-subjects experiment. This description reflects the conditions actually tested; it does not resolve the confounding among the visual attributes.

The opening-mechanism contrast compares hand-opening with tool-assisted opening across matched visual configurations. V2–V1 compares branded, printed natural cardboard with unbranded, unprinted natural cardboard. V3–V2 compares two branded presentations that differ jointly in printing approach and color. V3–V1 compares the corresponding complete configurations, with differences in logo presence, printing, and color. None of these visual contrasts separately identifies an independent effect of brand identity or printing technology.

We have revised the hypotheses, factor labels, and interpretation to reflect these boundaries. Section 6.3 explicitly acknowledges the design constraint and recommends future experiments that independently manipulate the visual attributes to distinguish their contributions.

Location: Section 3.2 and Table 2, pp. 58–62; Sections 2.3–2.5, pp. 47–56; Sections 5.2.2–5.2.3, pp. 84–88; Section 6.3, p. 97.

Q4 Insufficient reliability of the mediation analysis

The mediation analyses presented in Tables 4-10 require substantial reconsideration.

Secondly, the proposed mediators and purchase intention seem to be measured at the same evaluation episode, which complicates causal mediation claim.

There are also some numerical discrepancies in the reported results:

Table 4 shows an indirect effect of 0.164, whereas the corresponding text mentions 0.172.

The text before Table 8 states path a equal to 0.049, whereas Table 8 reports 1.049.

Some of the indirect effects are not consistent with the product of path coefficients shown in Tables 7 and 10.

This raises doubts about model specification and interpretations and reporting of those interpretations.

Response：Thank you for this detailed assessment. We have replaced the original mediation results with within-subject analyses using MEMORE 3.0, Model 1, in SPSS. Scores were averaged across visual configurations for the opening-mechanism contrast and across opening mechanisms for the visual contrasts. Differences were consistently coded as hand-opening minus tool-assisted opening, V2 minus V1, V3 minus V2, and V3 minus V1.

We have retained the single-mediator analyses and added four parallel mediation models. The opening-mechanism parallel model simultaneously includes perceived opening fluency, perceived product quality, and perceived brand image; each visual-configuration model includes perceived product quality and perceived brand image. Outcome regressions include mediator difference scores and mean-centered mediator-average terms. Coefficients are unstandardized, and each specific indirect effect is calculated as the product of its corresponding a and b coefficients.

Percentile bootstrap confidence intervals use 50,000 participant-level resamples, preserving paired observations, with seed 20261016. The nine single-mediator indirect effects and nine parallel-model specific indirect effects form separate Bonferroni-adjusted families, each using a confidence level of approximately 99.4%. These adjustments do not provide joint familywise error control across both families.

Tables 7–8 and the corresponding text now report the newly estimated results, replacing the earlier inconsistent coefficients. In the parallel models, the opening-mechanism indirect effect through perceived product quality met the adjusted criterion; the remaining specific indirect effects did not. The revised interpretation distinguishes single-mediator associations from specific indirect effects accounting for other mediators. Because the mediators and purchase intention were measured within the same evaluation episode, we describe statistical indirect associations rather than established causal or complete mediation.

Location: Section 5.2.3 and Tables 7–8, pp. 86–90; Discussion, pp. 91–94; Conclusions, p. 98.

Q5 Sample size, statistical power, and generalizability

The authors should provide a more specific power analysis, justify the precision of the estimates.

Also, limited demographic and educational diversity limits the generalizability of the findings to other cosmetics e-commerce population.

Response：Thank you for this comment. We have revised the sample-size justification in Section 4.1. Sample-size requirements were evaluated separately for the two main effects using the “ANOVA: Repeated measures, within factors” procedure in G*Power 3.1, based on participant-level scores averaged across the other factor. Assuming f = 0.25, α = 0.05, target power = 0.80, a repeated-measures correlation of 0.50, and ε = 1, the required sample sizes were 34 for opening mechanism and 28 for visual configuration. The final sample of 41 participants exceeded the larger estimate under these assumptions.

We explicitly limit this justification to the two main effects. It does not establish adequate power for interactions, indirect effects, or all multiplicity-adjusted comparisons. To communicate estimation precision, Table 6 reports mean differences with confidence intervals, and Tables 7–8 report bootstrap standard errors and confidence intervals for indirect effects. Table 5 also reports partial eta squared. These estimates and intervals make the magnitude and uncertainty of the findings explicit rather than treating sample size alone as evidence of sufficient precision.

Section 6.3 now acknowledges that convenience recruitment from Ningbo University and the surrounding community, together with limited demographic and educational diversity, restricts generalizability to broader cosmetic e-commerce consumer populations. We recommend replication with larger samples and more diverse demographic and educational backgrounds.

Location: Section 4.1, pp. 61–63; Tables 5–8, pp. 80–90; Section 6.3, sample-related limitation, p. 97.

Q6 Measurement validity and construct operationalization

Section 4.2 provides insufficient information about measurement items, scale adaptation, and psychometric properties.

Purchase intention is measured with a single item, whereas the other constructs are measured with adapted scales. The manuscript does not sufficiently validate the measurement instruments used.

Also, there is a need for the justification of the difference between perceptual processing fluency and physical ease of opening.

The authors should report the complete measurement items, construct operationalization, reliability statistics, and validation.

Response：Thank you for this comment. We have expanded Section 4.2 to specify the constructs, item counts, scoring procedures, response anchors, and adaptation sources. Appendix A, Table A1 provides English translations of all administered measurement items and their anchors.

We now consistently use “perceived opening fluency” for the subjective ease and effort involved in physically opening the packaging. Section 2.5 distinguishes this experience from information-processing fluency. Labroo and Lee provide a reference for assessing subjective processing ease, but their advertisement-processing measure is not treated as a validated measure of physical opening. Section 4.2 clarifies this distinction, and Table A1 reports the wording and anchors actually administered. The limitations acknowledge the single-item measure’s restricted coverage and need for further validation.

Purchase intention represents stated willingness, following the opening experience, to purchase products from the brand using the packaging. It is a prospective purchase intention informed by that experience, rather than an initial pre-experience judgment, observed purchasing behavior, or a measure specifically requiring a previous purchase from the brand.

Internal consistency was assessed separately within each condition using all 41 participants. Cronbach’s α ranged from 0.860 to 0.954 for perceived product quality and from 0.841 to 0.962 for perceived brand image; condition-specific results appear in Appendix A, Table A2. We acknowledge that these coefficients do not establish construct validity, discriminant validity, or measurement invariance. Purchase intention and perceived opening fluency are single-item measures, for which internal consistency cannot be assessed. Section 6.3 discusses their limited construct coverage and the need for further measurement development and validation.

Location: Section 2.5, pp. 49–53; Section 4.2, pp. 63–64; Appendix A, Tables A1–A2, pp. 106–107; Section 6.3, p. 97.

Q7 Inconsistency in the qualitative word association analysis

Section 5.1 mentions 738 initial associations and 732 valid associations, whereas the number of respondents mentioned is 40 despite the fact that the experiment involves 41 participants.

The manuscript should clarify whether there is any participant who was excluded from the qualitative analysis and what are the criteria for this exclusion.

Although the associations were independently coded by three researchers, the manuscript does not mention the measure of intercoder agreement before consensus discussion.

The authors should provide a detailed coding protocol, clarify the relationship between the initial and final numbers of associations and report appropriate measure of intercoder agreement.

Response：Thank you for identifying these discrepancies. We have corrected and clarified the response counts in Section 5.1. The previous figure of 738 represented the maximum number of response slots (41 participants × six conditions × three responses), rather than the number actually collected. There were 732 recorded responses. All 18 word-association responses from one participant were excluded because they were highly repetitive and mutually contradictory, leaving 714 responses from 40 participants for the qualitative analysis. This exclusion concerned the word-association responses; the quantitative analyses retained all 41 participants.

We have also clarified the distinction between the complete qualitative dataset and the frequency summary in Figure 6. The full retained dataset was used for semantic categorization and the 48-category analysis. For Figure 6, the frequency summary retained terms mentioned by at least four distinct participants, yielding 566 mentions across nine semantic dimensions. Lower-frequency terms were omitted from this figure, rather than excluded from the complete qualitative dataset. Visual perception accounted for 195 of these 566 mentions (34.5%); this percentage was not calculated using all 714 retained responses.

Section 5.1 now describes how the 444 distinct words or phrases were independently classified by three researchers, how disagreements were resolved through consensus to obtain 48 categories, and how these categories were subsequently organized into nine broader semantic dimensions.

Location: Sections 5.1–5.1.2, pp. 67–77; Table 3, pp. 69–70; Figures 5–6, pp. 71–73.

Q8 Contradictions between hypotheses, results, and interpretation

Several contradictions in the manuscript need correction.

In particular, the abstract states the positive impact of watermark printing on purchase intention, while Section 5.2.3 finds higher purchase intention in the case of offset printing.

Also, the printing hypothesis concerning purchase intention is designated as H3b in the theoretical section and H3c in the results section.

The H3a hypothesis described in Section 5.2.3 is formulated wrongly as a hypothesis concerning brand identity, not printing.

These contradictions should be removed through a careful audit of hypotheses and results.

Response：Thank you for identifying these inconsistencies. We have corrected the abstract to report higher purchase intention for V3, featuring offset printing with a white logo on a black background, than for V2, featuring watermark printing with a black logo on natural cardboard. This finding is interpreted as a difference between the tested visual configurations, rather than an independent effect of printing technology.

The H3 hypotheses and their corresponding results now use consistent numbering: H3a concerns perceived product quality, H3b perceived brand image, and H3c purchase intention, all for the V3–V2 comparison. Section 5.2.2 explicitly links the findings to these hypotheses. The estimated marginal mean differences favor V3 by 0.407 for perceived product quality (Bonferroni-adjusted p = 0.001), 0.654 for perceived brand image (Bonferroni-adjusted p < 0.001), and 0.500 for purchase intention (Bonferroni-adjusted p = 0.002). The erroneous description of H3a as a brand-identity hypothesis has been replaced by the correct visual-configuration comparison. The Conclusions likewise report the ordering V3 > V2 > V1 for these three outcomes.

Location: Abstract, pp. 38–39; Section 2.4 and H3a–H3c, pp. 49–50; Section 5.2.2, pp. 84–86; Table 6, pp. 82–84; Section 7, pp. 99–100.

Q9 Exaggeration of theoretical and practical implications

The distinction between the application and the theoretical advance should be clearer.

These implications should be deleted or at least stated as assumptions.

Response：Thank you for this comment. We agree that applying an existing framework in a new context should be distinguished from advancing or extending the theory itself. We have reframed the theoretical contribution as a contextual application of the S-O-R framework to cosmetic e-commerce packaging, examining the opening experience alongside visual presentation. We no longer claim to introduce perceptual fluency theory or establish a general theoretical extension. The Discussion also distinguishes observed differences in consumer evaluations from evidence for the proposed indirect associations, acknowledging both the supported opening-mechanism association through perceived product quality and the uncertainty surrounding the other specific pathways in the parallel models.

We have also narrowed the practical implications. Claims of reduced packaging costs and improved sustainability have been removed. Potential implications for shopping satisfaction, brand trust, and longer-term emotional connections are explicitly identified as requiring further investigation rather than presented as demonstrated outcomes. Practical recommendations are tied to the higher evaluations of hand-opening packaging and the branded, printed configurations tested, without generalizing these findings to independent effects of logo presence or printing technology.

Location: Discussion and Sections 6.1–6.2, pp. 94–97; Section 7, pp. 99–100.

Q10 Figures and tables

Response：Thank you for these suggestions. We have reorganized the quantitative tables to distinguish descriptive statistics (Table 4), repeated-measures ANOVAs (Table 5), estimated marginal mean comparisons (Table 6), single-mediator analyses (Table 7), and parallel mediation analyses (Table 8). Construct names, contrast directions, coefficients, and significance reporting have been aligned with the revised analyses. The table notes specify the relevant correction families and distinguish Holm-adjusted omnibus tests, Bonferroni-adjusted pairwise comparisons, and adjusted bootstrap intervals. Section 5.2.3 and the mediation table notes describe the included mediators, within-subject coding, and bootstrap settings.

We have also clarified the presentation and interpretation of Figures 5–7. The text accompanying Figure 5 describes differences in semantic profiles without interpreting central positions as neutral evaluations or proximity alone as a direct measure of association strength. For Figure 6, we distinguish the 566 retained mentions used for the frequency summary from the complete qualitative dataset of 714 responses and explicitly identify the percentage denominator. Figure 7 labels Dimension 1 and Dimension 2 and reports their explained inertia; its caption identifies the nine semantic dimensions and six packaging conditions, while the interpretation is presented in Section 5.1.2. These qualitative results are treated as exploratory descriptions rather than evidence establishing causal or mediation mechanisms.

Location: Sections 5.1–5.2, pp. 67–90; Figures 5–7, pp. 71–77; Tables 4–8 and their notes, pp. 78–90.

Q11 Accuracy and consistency of references

Response：Thank you for identifying these reference problems. We have removed the duplicate entries for Dodds et al. (1991), Hensler et al. (2015), Pousette et al. (2014), and Reber et al. (2004), retaining one entry for each publication. We have also standardized numerical citations and updated the reference numbering, eliminating the unnecessary year suffixes associated with these duplicate records.

In Section 2.4, we now distinguish technical research on printed-output characteristics from consumer research on packaging evaluations. Khadzhynova and Havenko (2020) and Mahajan and Bandyopadhyay (2020) are used to discuss print quality and production conditions, whereas Bone and France (2001) and Orth and Malkewitz (2008) inform the discussion of product beliefs and brand impressions. Technical printing evidence is not presented as direct evidence of consumer responses or as proof that printing technology independently caused the observed differences.

Location: Section 2.4, pp. 49–50; in-text citations throughout the manuscript; References, pp. 101–106.

Q12 English language and editorial consistency

The terminology should be consistent in the manuscript in terms of opening approaches, printing approaches, perceived fluency and purchase intention.

Response：Thank you for these comments. We have revised the theoretical, results, discussion, and conclusion sections to improve clarity, reduce repetition, and correct grammatical and typographical errors. The revised terminology consistently distinguishes “opening mechanism” from “visual configuration” and uses “perceived opening fluency,” “perceived product quality,” “perceived brand image,” and “purchase intention” for the measured constructs.

The previously repeated accounts of printing-related results have been replaced by a consolidated presentation of visual-configuration comparisons in Section 5.2.2. Section 5.2.3 now reports the single-mediator and parallel mediation analyses, distinguishing indirect associations from differences in outcome ratings. The Discussion likewise considers brand information and printed presentation together at the configuration level, avoiding repetitive or inconsistent attribution of the findings to individual visual attributes. Corresponding terminology has been aligned across the hypotheses, results, table headings and notes, Discussion, and Conclusions.

Location: Throughout the manuscript, pp. 38–106, particularly Section 2, pp. 43–53; Section 5.2, pp. 77–89; Sections 6–7, pp. 91–100; Table 2, pp. 61–62, and Tables 4–8, pp. 78–90.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The recommendations have been incorporated by the authors in the manuscript adequately.

Author Response

Comment 1: The recommendations have been incorporated by the authors in the manuscript adequately.
Response 1: Thank you for reviewing our revised manuscript and for acknowledging that your recommendations have been adequately addressed. We appreciate your constructive feedback, which has helped us improve the manuscript.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Substantive amendments have been made by the authors to the manuscript, and the response to my concerns from the previous review is constructive. The revisions indicate clear improvements in methodology, statistics, and interpretation of the findings.

First of all, changing the statistics from those in the original paper to repeated-measures ANOVA addresses the dependence among the observations. Within-subject mediation analysis with parallel models and correction for multiple tests is another improvement made by the authors. Moreover, I would like to acknowledge the fact that printing techniques, branding, and color schemes were not manipulated independently. Therefore, the interpretation of the results should be limited to the experimental design only.

The conclusions in the revised paper are largely consistent with the findings, and the implications of the results are formulated more carefully. However, several issues need to be discussed before the publication of the paper.

Validity of constructs: Since the experiment includes a fictitious brand, the meaning of brand image items measuring leadership in the market, experience in the field, and customer focus needs to be re-examined as they seem to indicate impressions generated by packaging and not brand evaluations based on knowledge about the brand.

Inter-rater agreement of qualitative coding: The authors should acknowledge that pre-consensus coding of participants' word associations was not saved, thus inter-rater agreement cannot be calculated. The criteria that led to discarding the word associations of one participant also should be elaborated.

Experimental procedure: It needs to be clarified whether the packages included products inside or were empty and whether participants understood which brand was evaluated when packages lacked any brand-related information.

Statistical interpretation: It is necessary to maintain the cautious interpretation of the mediation results due to the relatively small number of participants and the simultaneous measurement of the mediators and purchase intention. Providing anonymized data files and statistics scripts would increase the replicability of the study, provided that it is ethically possible.

Tables and presentation: There are several minor inaccuracies that need to be corrected. It should be specified what kind of brand identity information corresponds to the categories in Table 1, because the surveyed brands are identified with their logos. In Table 4, there seems to be a typo ("5s.30"). Also, Table 2 can be better organized.

In general, the authors have sufficiently addressed the methodological and analytical concerns raised in the previous review. The remaining issues are relatively minor and can be solved without making any major changes in the experimental design and statistical analysis.

I, therefore, recommend accepting this manuscript after minor revision.

Comments on the Quality of English Language

The quality of the English language has been greatly enhanced compared to the previous version. However, it would be prudent for some linguistic editing to be conducted to address grammar mistakes, awkward sentence construction, and problems concerning punctuation and formatting. It is crucial that special consideration be given to the consistency of statistical terms, table captions, and other technical information.

Author Response

Q1 Validity of constructs: Since the experiment includes a fictitious brand, the meaning of brand image items measuring leadership in the market, experience in the field, and customer focus needs to be re-examined as they seem to indicate impressions generated by packaging and not brand evaluations based on knowledge about the brand.

Response: Thank you for highlighting this issue. We agree that, for a fictitious brand, ratings of market leadership, industry experience, and customer orientation cannot be interpreted as evaluations grounded in prior brand knowledge or evidence of actual brand performance. We have revised the Discussion to interpret these ratings more cautiously as initial brand impressions inferred from the packaging in the experimental context. We have also added a limitation acknowledging that this interpretation does not establish the construct validity of the measure and that the suitability of these items for initial exposure to a fictitious brand remains uncertain. Future research is therefore encouraged to use measures better suited to this context and further assess their construct validity. The administered items, the perceived brand image variable name, and the statistical results remain unchanged.

Location: Section 6, Discussion, p. 63; Section 6.3, Limitations and Future Research, p. 69.

Q2 Inter-rater agreement of qualitative coding: The authors should acknowledge that pre-consensus coding of participants' word associations was not saved, thus inter-rater agreement cannot be calculated. The criteria that led to discarding the word associations of one participant also should be elaborated.

Response: Thank you for this comment. We have clarified the coding-record limitation in Section 5.1.1. Although independent coding files were retained, complete item-level pre-consensus records were not available in a form that allowed the three researchers’ classifications to be reliably aligned. Consequently, pre-consensus intercoder agreement could not be calculated. We acknowledge that this limits the assessment of coding reproducibility.

We have also elaborated the exclusion rationale in Section 5.1. Before data organization and coding, all 18 word associations from one participant were excluded because of their highly repetitive pattern across packaging conditions. The participant provided the identical triplet “convenient, troublesome, happy” for the first two conditions, and “troublesome” and “happy” accounted for 11 of the 18 responses. The qualitative analysis therefore included 714 responses from 40 participants. This exclusion applied only to the word association analysis; all 41 participants were retained in the questionnaire analyses.

Location: Section 5.1, p. 36; Section 5.1.1, p. 37.

Q3 Experimental procedure: It needs to be clarified whether the packages included products inside or were empty and whether participants understood which brand was evaluated when packages lacked any brand-related information.

Response: Thank you for requesting this clarification. Section 4.3 now specifies that each package contained an identical 50 g calibration weight rather than an actual cosmetic product, and that the weight was visible after opening. All participants received standardized verbal instructions identifying the six packaging conditions as belonging to the same fictitious brand, “You Face.” Participants therefore knew which brand they were evaluating in the V1 conditions, although no logo or other brand information appeared on the packaging.

Location: Section 4.3, Procedure, pp. 35–36.

Q4 Statistical interpretation: It is necessary to maintain the cautious interpretation of the mediation results due to the relatively small number of participants and the simultaneous measurement of the mediators and purchase intention. Providing anonymized data files and statistics scripts would increase the replicability of the study, provided that it is ethically possible.

Response: Thank you for this comment. We have added a paragraph to Section 6.3 acknowledging that the relatively small sample and the assessment of the proposed mediators and purchase intention in the same questionnaire after each package was opened call for cautious interpretation of the mediation findings. We clarify that the estimated indirect effects represent statistical associations and do not establish causal mediation. To facilitate reproducibility, anonymized data, the SPSS analysis syntax, and the Python script for the presentation-position analysis are available upon request.

Location: Section 6.3, Limitations and Future Research, p. 69.

Q5 Tables and presentation: There are several minor inaccuracies that need to be corrected. It should be specified what kind of brand identity information corresponds to the categories in Table 1, because the surveyed brands are identified with their logos. In Table 4, there seems to be a typo ("5s.30"). Also, Table 2 can be better organized.

Response: Thank you for identifying these presentation issues. In Table 1, we have replaced “Brand Identity Information” with “Brand logo on the packaging exterior” and added a note distinguishing the lettering of the brand logo itself from its presence on the surveyed e-commerce packaging. In Table 2, we have grouped the conditions under the two opening mechanisms and merged repeated opening-mechanism labels to make the 2 × 3 design clearer. In Table 4, we have corrected “5s.30” to “5.30” for perceived brand image in Group 5. These changes clarify the presentation without altering the classifications, experimental conditions, or underlying statistical results.

Location: Table 1, pp. 26–27; Table 2, pp. 30–31; Table 4, p. 48.

Q6 Comments on the Quality of English Language

Response: Thank you for this suggestion. We have made further language and editorial corrections to improve clarity and consistency, including corrections to spacing, punctuation, and typographical errors. We have also clarified table headings and checked the consistency of terminology used for the experimental conditions and measured constructs.

Location: Throughout the revised manuscript.

Author Response File: Author Response.docx