Review Reports - Managing Food Leftovers in Polish Households in Terms of the Food Waste Hierarchy

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper presents a well-structured study on Polish household practices in managing food leftovers, aligned with the food waste hierarchy. The dataset is large (n = 1115), representative, and the analytical methods (ANOVA, logistic regression) are appropriate. The paper contributes meaningfully to sustainability and consumer behavior research.

General comments

Please add to the abstract the year of data collection (i.e., 2019) and justify its continued relevance post-COVID. It is not enough to mention that introduction.

Line 25: mention food waste and then (FW) as it is the first time to be mentioned

Line 34: what does number 1 mean after estimates?! Is it a reference?

Line41: I believe the reference here is for the previous part. Please correct the references and add the one of Eurostat

The introduction is too long. Condense the literature review by merging overlapping references on household FW determinants (lines 30–100). Additionally, the introduction does not explain the food waste hierarchy as an analytical lens. Please enhance this part.

Clarify dependent and independent variables in regression, Table 6 lists them, but they should be briefly introduced in the text before the model.

Mention data validation (missing values handling, outlier removal, or weighting).

Improve readability of some tables such as tables (2–4) have small text and redundant decimal points.

Tables 3 and 4 specify under the tables what does the asterisk and small letters mean. Use the small letters in the table as superscript.

Most references are missing the DOI

Author Response

Dear Reviewer,

thank you very much for the review and all the helpful comments — they are a great support for us as I continue working on the revisions.

Comment 1: Please add to the abstract the year of data collection (i.e., 2019) and justify its continued relevance post-COVID. It is not enough to mention that introduction

Response 1. We have added the year of data collection to the abstract and addressed it in the Discussion.

Comment 2: Line 25: mention food waste and then (FW) as it is the first time to be mentioned.

Response 2: We improved.

Comment 3: Line 34: what does number 1 mean after estimates?! Is it a reference?

Response 3: We removed number 1.

Comment 4: Line41: I believe the reference here is for the previous part. Please correct the references and add the one of Eurostat.

Response 4: The source given is correct.

Comment 5: The introduction is too long. Condense the literature review by merging overlapping references on household FW determinants (lines 30–100). Additionally, the introduction does not explain the food waste hierarchy as an analytical lens. Please enhance this part.

Response 5: Thank you for pointing this out. We shortened the literature review and combined information on the same determinants. We have explained the food waste hierarchy as an analytical lens (lines 54-56).

Comment 6: Improve readability of some tables such as tables (2–4) have small text and redundant decimal points.

Response 6: Thank you for this valuable comment. The font size used in Tables 2–4 is 10 pt, which is consistent with the journal’s template. To maintain consistency in the presentation of numerical data throughout the manuscript, we retained two decimal places for all quantitative variables. However, we reviewed the tables to ensure clarity and readability and confirmed that the formatting remains uniform and easy to interpret. If necessary, we are prepared to make further adjustments to the tables in accordance with the journal’s production requirements during the proof stage.

Comment 7: Tables 3 and 4 specify under the tables what does the asterisk and small letters mean. Use the small letters in the table as superscript.

Response 7: We improved.

Comment 8: Most references are missing the DOI.

Response 8: We added missing DOI.

Reviewer 2 Report

Comments and Suggestions for Authors

1. Summary of the Manuscript and Key Contributions

The paper, “Managing food leftovers in Polish households in terms of the food waste hierarchy,” investigates behavioral and sociodemographic factors influencing how Polish households manage uneaten meals. It presents findings from a nationwide survey of 1,115 adults, providing empirical evidence on how factors such as age, residence, education, household composition, and income affect food leftover management practices. The study is one of the few in Central Europe to apply quantitative statistical modeling (ANOVA and logistic regression) to identify determinants of food disposal behaviors, contributing to household food waste management literature and policy recommendations relevant to SDG 12.3 (Responsible Consumption and Production).

2. Detailed Evaluation of the Methodology, Analyses, and Conclusions

Methodology

Survey Design: The study employed computer-assisted personal interviews (CAPI) with structured questionnaires divided into behavioral and demographic sections. Respondents rated the frequency of six leftover management methods using a five-point Likert scale (“always” to “never”).
Variables: Independent variables included shopping habits, perception of product features, food handling, meal preparation behavior, and attitudes toward food waste. The dependent variable was throwing leftovers into the waste container.
Statistical Approach:
- Descriptive statistics (mean, median, mode, SD) to summarize disposal behaviors.
- One-way ANOVA to test significant differences in management practices across demographic groups (gender, age, residence, education, household size, income, etc.).
- LSD post-hoc tests to determine homogeneous groups.
- Logistic regression modeling to identify predictors of throwing away leftovers, using p < 0.25 as a selection threshold (following Hosmer & Lemeshow’s recommendation).
Strength: The multilevel approach (ANOVA + regression) enhances analytical rigor.
Limitation: Behavioral data were self-reported, increasing the likelihood of social desirability bias; the authors note this, but could have complemented it with direct observations or diaries.

Analysis

Descriptive Insights: Only 25% of respondents “always” or “usually” reused leftovers; disposal into waste containers ranked third, showing a lack of adherence to the food waste hierarchy.
Socio-Demographic Effects: Rural residents more frequently fed animals or composted leftovers, while urban and younger consumers preferred disposal.
Statistical Findings:
- ANOVA results confirmed significant effects (p < 0.05) of place of residence, age, and household characteristics on disposal behavior.
- Regression analysis revealed that purchase frequency of ready-made meals and perceived importance of storage conditions were the strongest predictors of wasteful behavior.
- Price sensitivity and attention to storage correlated with less wasteful practices.
Interpretation: The model effectively highlights consumer behavior segments with high waste risk, particularly younger urban consumers with lower food management awareness.

Conclusions

The paper concludes that Polish households do not adhere to the food waste hierarchy, with disposal still ranking high among common behaviors.
Education and awareness campaigns targeting youth and urban populations are emphasized as policy priorities.
The study provides a data-driven foundation for sustainable food waste management interventions.
Limitations include the lack of behavioral validation and the omission of post-pandemic consumer trends.

3. Constructive Feedback for Improvement

Methodological Transparency:
- Provide more detail on questionnaire validation, reliability (Cronbach’s α), and sampling procedures (response rate, regional distribution).
- Explicitly state ethical approval and consent processes, even if data collection was conducted by a professional agency.
Statistical Reporting:
- Include effect sizes (η², OR with confidence intervals) to better interpret the strength of observed relationships.
- Clarify handling of “hard to say” responses and potential non-normality in ANOVA assumptions.
Discussion Depth:
- Expand on behavioral theories (e.g., Theory of Planned Behavior) to contextualize findings.
- Compare results with post-2020 European studies on consumer waste behaviors (e.g., pandemic-related changes in meal planning).
Conclusions and Policy Implications:
- Integrate quantitative outcomes with concrete policy recommendations (e.g., incentive programs for composting or food donation).
- Reframe the conclusion to emphasize contribution to the circular economy and SDG 12.3 implementation pathways.

Comments on the Quality of English Language

Consider shortening long sentences for better readability.
Replace phrases like “it should be stated that” with more direct expressions (e.g., “This suggests…”).
Ensure consistent use of terms (e.g., “FW hierarchy” vs. “food waste hierarchy”).
Use formal tone:
- Replace “people” → “consumers”
- Replace “threw away” → “disposed of”
- Replace “used” → “utilized” or “repurposed”
Check for grammar consistency (past tense in Methods, present tense in Discussion).

Author Response

Dear Reviewer,

thank you very much for the review and all the helpful comments — they are a great support for us as I continue working on the revisions.

Comment 1: Provide more detail on questionnaire validation, reliability (Cronbach’s α), and sampling procedures (response rate, regional distribution).

Response 1: The presented analysis used individual questions from a broader research questionnaire. This information has been included in the “Data collection and questionnaire” section. Therefore, Cronbach's alpha coefficient was not calculated for the individual explanatory variables used in the present analysis, as they are fragments of broader questions in the original questionnaire. It should be emphasised, however, that in the original questionnaire, reliability was satisfactory (above 0.7). For example, the question about preparing a shopping list was grouped with two other questions about consumers' preparation before going shopping, and the Cronbach's alpha value for this group of questions was 0.785.

The regional distribution of the study participants has also been described in the Materials and Methods section, and an illustration has been added. Please refer to lines 159.–162 and Figure 1.

We do not have information on the proportion of respondents who declined to participate in the survey. However, the Methods section has been supplemented with a detailed description of the interviewer route procedure, which ensured an appropriate and representative sampling process. Please refer to lines 154.–160.

Comment 2: Explicitly state ethical approval and consent processes, even if data collection was conducted by a professional agency.

Response 2: Thank you for pointing this out. The section Materials and Methods has been supplemented. Please refer to lines 145–152.

Comment 3: Include effect sizes (η², OR with confidence intervals) to better interpret the strength of observed relationships.

Response 3: The effect size (η²p) has been calculated and included in the analysis. The results are presented in Table. Please refer to lines 214–219.

Comment 4: Clarify handling of “hard to say” responses and potential non-normality in ANOVA assumptions.

Response 4: In the ANOVA analysis, the response option “hard to say” was treated as missing data and was therefore excluded from the calculations. The study was conducted only for the responses within the 1–5 scale, ensuring consistency in the level of measurement and interpretation of the results. According to the statistical literature (Stanisz A. (2006): An accessible statistics course using STATISTICA PL. Volume 1: Basic statistics. StatSoft Polska.), with a large sample size, which was the case in this study, ANOVA analysis is characterised by resistance to moderate deviations from the normal distribution. Many authors argue that parametric tests, including ANOVA, maintain the correctness of inference for large samples, even if the distribution deviates from the normal one.

Comment 5: Expand on behavioral theories (e.g., Theory of Planned Behavior) to contextualize findings.

Response 5: We developed a Theory of Planned Behavior (lines 87-96).

Comment 6: Compare results with post-2020 European studies on consumer waste behaviors (e.g., pandemic-related changes in meal planning).

Response 6: Thank you for pointing this out. We have supplemented the literature with new items published in the last three years (e.g. 29-34).

Comment 7: Reframe the conclusion to emphasize contribution to the circular economy and SDG 12.3 implementation pathways.

Response 7: We have reframed the conclusions to emphasize our contribution to the circular economy and the implementation of SDG 12.3.

Comment 8: Consider shortening long sentences for better readability.

Replace phrases like “it should be stated that” with more direct expressions (e.g., “This suggests…”).
Ensure consistent use of terms (e.g., “FW hierarchy” vs. “food waste hierarchy”).
Use formal tone:
Replace “people” → “consumers”
Replace “threw away” → “disposed of”
Replace “used” → “utilized” or “repurposed”
Check for grammar consistency (past tense in Methods, present tense in Discussion

Response 8: Thank you for this valuable comment. The article was verified by a native speaker.

Reviewer 3 Report

Comments and Suggestions for Authors

The introduction states that the study aims to examine "how Polish households manage food leftovers" and analyzes socio-economic factors.This resembles a broad research area rather than a specific, testable research question.
The core dependent variable for the logistic regression is the behavior of "throwing uneaten meals into a waste container."However, the merging of response categories into "always/usually/sometimes" (Y=1) versus "rarely/never/hard to say" (Y=0) is highly problematic. This grouping conflates medium-frequency ("sometimes") with high-frequency ("always," "usually") behaviors, severely blurring the definition of the behavior itself. Consequently, the model attempts to predict an ill-defined behavioral composite.
The final regression model includes only a few variables, while many factors significant in the univariate analysis (as mentioned in the introduction and results) were excluded.The authors provide no justification for why these variables were omitted or for the rationale behind the selection of the current variables for the final model.
As shown in Table 7, the logistic regression model's predictive accuracy for the "throwing into container" behavior (Y=1) is only 34% (125/367), which is even lower than random guessing.This indicates that the model is entirely ineffective at distinguishing the target behavior, casting serious doubt on the practical utility of its identified "significant influencing factors" (e.g., frequency of purchasing frozen meals).
There are multiple instances of erroneous or contradictory data interpretation in the Results and Discussion sections.For example, in Section 3.2, it is stated: "Consumers over 45, including those over 60, were much less likely to declare throwing unused meals to the container compared to young consumers aged 25-34." However, data in Table 4 show that the mean values for the 45-59 and over-60 age groups (3.69 and 3.80, respectively) are both higher than that of the 25-34 age group (3.42). Given that a higher mean indicates a greater tendency to discard, this textual description directly contradicts the presented data.
In Section 3.3 and Table 5, it is claimed that "in the R_Yes group, about 3/4 of respondents declared that, on average, they purchase chilled ready-made meals every other day."However, the percentage for the "every other day" frequency in Table 5 represents the proportion within that specific response category, not the proportion of the entire R_Yes group. This constitutes a fundamental misinterpretation of the crosstabulation data.
Another contradiction is found in Section 3.2: "Households without children also threw such products into the waste container less often, compared primarily to households with one child."Table 4 data show that the mean for households with no children (3.71) is higher than that for households with one child (3.42), again rendering the textual description inconsistent with the data.
The introduction dedicates substantial space to reviewing the global context of food waste and its influencing factors (e.g., planning, storage).However, the core of this study is the "choice of disposal pathways for leftovers." The introduction fails to effectively connect the reviewed literature to this specific behavior, resulting in a disconnection between the literature review and the subsequent analysis.
The data were collected in 2019, prior to the COVID-19 pandemic, which significantly altered global food purchasing and consumption patterns.The authors only briefly mention this in the introduction's conclusion but fail to adequately discuss this limitation in the Discussion and Conclusion sections regarding its potential impact on the generalizability and current relevance of the findings.
The reference list of the manuscript has obvious timeliness issues and fails to fully include important literature published in the past three years. This weakens the academic cutting-edge nature of the research and the depth of the literature review.

Author Response

Dear Reviewer,

thank you very much for the review and all the helpful comments — they are a great support for us as I continue working on the revisions.

Comment 1: The introduction states that the study aims to examine "how Polish households manage food leftovers" and analyzes socio-economic factors. This resembles a broad research area rather than a specific, testable research question.

Response 1: We improved. We have removed statements that are too general.

Comment 2: The core dependent variable for the logistic regression is the behavior of "throwing uneaten meals into a waste container. "However, the merging of response categories into "always/usually/sometimes" (Y=1) versus "rarely/never/hard to say" (Y=0) is highly problematic. This grouping conflates medium-frequency ("sometimes") with high-frequency ("always," "usually") behaviors, severely blurring the definition of the behavior itself. Consequently, the model attempts to predict an ill-defined behavioral composite.

Response 2: The regression analysis employed required a dichotomous dependent variable, which justified the need to aggregate the response categories from the questionnaire. The study aimed to identify individuals who practice this behaviour, namely "throwing uneaten food into the trash," as opposed to those who do not practice it or do it only very occasionally. In this approach, the response "sometimes" was classified alongside the responses "always" and "usually," as it indicates the actual occurrence of the analysed behaviour. However, the reactions "rarely" and "never" were treated as indicating a lack of a persistent habit of throwing away leftovers, and the response "difficult to say" was assigned to the same category due to the lack of a stable behavioural pattern.

An explanation of this grouping approach has been provided in the text in Section 2.3. Data analysis.

Comment 3: The final regression model includes only a few variables, while many factors significant in the univariate analysis (as mentioned in the introduction and results) were excluded. The authors provide no justification for why these variables were omitted or for the rationale behind the selection of the current variables for the final model.

Response 3: We describe the method of selecting variables for the regression model in detail in Section 2.3 of the Methodology—data analysis, where we present the criteria for including and excluding predictors . In the revised version of the manuscript, we also emphasize this process again in the Results section, where we clarify that we omitted some variables from the final model.

Comment 4: As shown in Table 7, the logistic regression model's predictive accuracy for the "throwing into container" behavior (Y=1) is only 34% (125/367), which is even lower than random guessing. This indicates that the model is entirely ineffective at distinguishing the target behavior, casting serious doubt on the practical utility of its identified "significant influencing factors" (e.g., frequency of purchasing frozen meals).

Response 4: We acknowledge the low predictive accuracy of the logistic regression model for the "bin throwing" behaviour (Y=1), which is clearly stated in the manuscript. The goal of constructing the logistic regression model was to identify variables significantly associated with the likelihood of throwing uneaten food into the waste container. In this context, the interpretation of the regression coefficients remains valuable, as it indicates the direction and strength of the relationship between the explanatory variables and the behaviour, regardless of the model's ability to correctly classify individual cases. This model served as a complement to the previously conducted analyses, rather than its main predictive component.

Comment 5: There are multiple instances of erroneous or contradictory data interpretation in the Results and Discussion sections. For example, in Section 3.2, it is stated: "Consumers over 45, including those over 60, were much less likely to declare throwing unused meals to the container compared to young consumers aged 25-34." However, data in Table 4 show that the mean values for the 45-59 and over-60 age groups (3.69 and 3.80, respectively) are both higher than that of the 25-34 age group (3.42). Given that a higher mean indicates a greater tendency to discard, this textual description directly contradicts the presented data.

Response 5: The interpretation in Section 3.2 is consistent with the data in Table 4. We would like to clarify that the response scale used in this study was reversed, where 1 = always, 2 = usually, 3 = sometimes, 4 = rarely, and 5 = never. Therefore, higher mean values indicate a lower tendency to discard uneaten meals, not a higher one. On this basis, consumers aged 45–59 (mean value = 3.69) and 60+ (mean value = 3.80) indeed reported discarding uneaten meals less frequently than respondents aged 25–34 (mean value = 3.42).

For clarity, the response scale has been reiterated below the table to facilitate the correct interpretation of the results.

Comment 6: In Section 3.3 and Table 5, it is claimed that "in the R_Yes group, about 3/4 of respondents declared that, on average, they purchase chilled ready-made meals every other day."However, the percentage for the "every other day" frequency in Table 5 represents the proportion within that specific response category, not the proportion of the entire R_Yes group. This constitutes a fundamental misinterpretation of the crosstabulation data.

Response 6: Thank you very much for this valuable remark. The values in the table have been corrected so that they now show the responses within the compared groups, i.e. respondents who dispose of uneaten meals in a waste bin and those who do not. The descriptions have been adjusted accordingly. (Please refer to lines 318 – 323).

Comment 7: Another contradiction is found in Section 3.2: "Households without children also threw such products into the waste container less often, compared primarily to households with one child."Table 4 data show that the mean for households with no children (3.71) is higher than that for households with one child (3.42), again rendering the textual description inconsistent with the data.

Response 7: This point is related to the earlier comment regarding the interpretation of the response scale. The interpretation in Section 3.2 is consistent with the data in Table 4. As clarified previously, the response scale used in this study was reversed, where 1 = always, 2 = usually, 3 = sometimes, 4 = rarely, and 5 = never. Therefore, higher mean values indicate a lower tendency to discard uneaten meals, rather than a higher one.

Comment 8: The introduction dedicates substantial space to reviewing the global context of food waste and its influencing factors (e.g., planning, storage).However, the core of this study is the "choice of disposal pathways for leftovers." The introduction fails to effectively connect the reviewed literature to this specific behavior, resulting in a disconnection between the literature review and the subsequent analysis.

Response 8: We have made changes to ensure that the introduction effectively connects the analysed literature with the research goal. 98-105

Comment 9: The data were collected in 2019, prior to the COVID-19 pandemic, which significantly altered global food purchasing and consumption patterns. The authors only briefly mention this in the introduction's conclusion but fail to adequately discuss this limitation in the Discussion and Conclusion sections regarding its potential impact on the generalizability and current relevance of the findings.

Response 9: Issues related to data collection in 2019, i.e., before the COVID-19 pandemic, and the possibility of generalising them and using them for comparisons with data collected after the pandemic are discussed in the Discussion chapter.

Comment 10: The reference list of the manuscript has obvious timeliness issues and fails to fully include important literature published in the past three years. This weakens the academic cutting-edge nature of the research and the depth of the literature review.

Response 10: We have updated the literature list with current titles (e.g. 29-34, 73-74).