Sustainable Fruit Consumption: The Influence of Color, Shape and Damage on Consumer Sensory Perception and Liking of Different Apples
Round 1
Reviewer 1 Report
The authors present a paper that is succinct and has relevance for the journal – the authors assess consumer liking and sensory appraisal of apples presented with combinations of optimal and sub-optimal colour, shape and damage characteristics (presented as images), in the context of understanding food waste from aesthetic differences in a fresh-produce. Although the content is relevant, the paper is not suitable for publication in its current form. More information on the methods, study design and statistical analyses is required in order to properly assess the quality of the research. Please see my specific comments below. I would be happy to re-consider the manuscript once these clarifications have been made.
Optimal/Suboptimal characteristics – The damage and shape scales seem appropriate but it is unclear why green was considered a suboptimal colour? Apples are typically red or green, so it is possible that the colour was not viewed as sub-optimal. I would consider brown or dull colouration a suboptimal colour, not green. Can the authors explain their choice here?
Figure 1 – This figure is quite nice. I am not sure if this is just the PDF I received, but the images of the apple should be in colour, not greyscale.
Methods – More detail is required to understand the study design.
Firstly, there appears to be 8 picture combinations paired with 8 different apple-samples… does this mean that there were (8 x 8) 64 combinations tasted and rated by all participants? Please clarify.
Secondly, how did these 8 samples of apples differ? Were they different types of apples? Different colours? How were they selected? What quantity (g) was served as a sample and how were they prepared? skin-on?
Thirdly, did all participants evaluate all sample combinations? Between-group comparisons are mentioned on line 102-103, but it is not clear if a between-groups design was employed (and how), or if this was a completed repeated-measures/within-groups study. The study design needs to be fully described.
Finally, the participants were not trained panellists, but it is not clear whether any descriptors were provided for each of the 13 attributes assessed. Please add this.
Statistical Analyses - The analysis plan requires clarification in order to understand the study design and whether the data have been treated correctly. For example, the authors state that Tukey-Kramer tests were used alongside 1-way ANOVAs, and that 14 tests were completed in total. However, it is unclear to me which variables (both independent and dependent variables) are assessed in these analyses? There are 8 pictures, 8 apple samples and 14 liking/sensory outcomes. Does this mean that separate within-groups ANOVAs with ‘picture’ as a factor (8 levels) were conducted for each of the 14 outcome attributes, and the data were collapsed across the 8 apple samples? Please clarify in the text. Rather than assessing the different apple-attributes (colour/damage/shape) within one variable, these could have been included as three different Factors (2 levels each – optimal vs suboptimal) within the same statistical models. Why was this not done?
Another query is that it is not clear whether any correction for multiple comparisons was applied. If it was not, it is possible that the significant differences reported may not remain significant. Please add in this detail and justify why corrections were not applied, if they were not.
Finally, the authors state that between-groups differences were assessed for liking and the 13 different attributes, but it is not clear how/why this can be as I was under the impression that this was a repeated-measures designed. What variables were assessed between-groups? Please clarify.
Results – The results of the statistical analyses (e.g. actual f-values and p-values) should be reported or summarised where possible, even for non-significant findings. Currently the authors primarily describe the results in text and it is unclear which analyses are being reported where and how ‘singificant’ and ‘non-significant’ these differences were without any indication of the relevant f-values and p-values.
Table 1 legend – P = 0.05 should be P > 0.05?
Author Response
Review 1,
Dear Reviewer, thanks for comments. We have taken them all into account and revised the manuscript. See below text in red for amendments.
The authors present a paper that is succinct and has relevance for the journal – the authors assess consumer liking and sensory appraisal of apples presented with combinations of optimal and sub-optimal colour, shape and damage characteristics (presented as images), in the context of understanding food waste from aesthetic differences in a fresh-produce. Although the content is relevant, the paper is not suitable for publication in its current form. More information on the methods, study design and statistical analyses is required in order to properly assess the quality of the research. Please see my specific comments below. I would be happy to re-consider the manuscript once these clarifications have been made.
Optimal/Suboptimal characteristics – The damage and shape scales seem appropriate but it is unclear why green was considered a suboptimal colour? Apples are typically red or green, so it is possible that the colour was not viewed as sub-optimal. I would consider brown or dull colouration a suboptimal colour, not green. Can the authors explain their choice here?
Thank you for notifying us! This is absolutely right. The subject is discussed in the discussion paragraph. There is, however, a need to clarify the red-green color already in the introduction. In line 58-59 the following sentence is thus added: It has further been shown that red apples are more liked than green ones, and that red colored apples may be seen as more optimal than for example green.
Figure 1 – This figure is quite nice. I am not sure if this is just the PDF I received, but the images of the apple should be in colour, not greyscale.
The figure is in color.
Methods – More detail is required to understand the study design.
Firstly, there appears to be 8 picture combinations paired with 8 different apple-samples… does this mean that there were (8 x 8) 64 combinations tasted and rated by all participants? Please clarify.
A clarification is certainly needed to avoid any misunderstanding. A table showing the eight samples in the cubic design is added (Table 1). The text is clarified and now reads (line 89-94): The study was based on a cubic full factorial design of visual appearance (color, shape and damage) presented in a total of eight samples, each consisting of a combination of pictures describing the optimality-suboptimality. The samples were ranging from optimal to sub-optimal in all three dimensions (based on combinations of color (red-to-green), shape (normal-to-odd) and damage (none-to-damage). The design is presented in Table 1 and visualized in Figure 1. The picture combinations for each sample were combined with the tasting of eight randomly selected apples.
Secondly, how did these 8 samples of apples differ? Were they different types of apples? Different colours? How were they selected? What quantity (g) was served as a sample and how were they prepared? skin-on?
The eight categories/samples were 1. the pictures (categories) in the design and 2 taste samples. The pictures differed according to color, shape and damage. The tasting samples were random apples of different cultivars, all grown in Sweden. The text in the material and method section has been rewritten to better explain the test procedure and the included samples.
Thirdly, did all participants evaluate all sample combinations? Between-group comparisons are mentioned on line 102-103, but it is not clear if a between-groups design was employed (and how), or if this was a completed repeated-measures/within-groups study. The study design needs to be fully described.
All consumers evaluated the same samples. This is clarified in line 105. Between group comparison is due to liking and part of the statistical evaluation. The text in the statistical evaluation is changed to clarify this. Please see below (Statistical Analyses).
Finally, the participants were not trained panellists, but it is not clear whether any descriptors were provided for each of the 13 attributes assessed. Please add this.
The questionnaire was self-instructive with no descriptors of the attributes. This is added in line 104.
Statistical Analyses - The analysis plan requires clarification in order to understand the study design and whether the data have been treated correctly. For example, the authors state that Tukey-Kramer tests were used alongside 1-way ANOVAs, and that 14 tests were completed in total. However, it is unclear to me which variables (both independent and dependent variables) are assessed in these analyses? There are 8 pictures, 8 apple samples and 14 liking/sensory outcomes. Does this mean that separate within-groups ANOVAs with ‘picture’ as a factor (8 levels) were conducted for each of the 14 outcome attributes, and the data were collapsed across the 8 apple samples? Please clarify in the text. Rather than assessing the different apple-attributes (colour/damage/shape) within one variable, these could have been included as three different Factors (2 levels each – optimal vs suboptimal) within the same statistical models. Why was this not done?
We performed a one-way analysis using 8 groups, and we performed this analysis first for liking and then for all 13 perceived sensory attributes. It is corrected that we could have used a three-way analysis with two layers per grouping variable; however, given the small number of groups (8) we felt that it is actually more clear to look at the “whole” grouping directly without adding the complexity of n-way ANOVA. We did confirm that the differences and significance of differences between groups remains exactly the same, and since this is what we aim to study (rather than the importance of pure vs interaction terms), we decided to settle for the simplest ANOVA analysis. We added a comment on this in the Statistical evaluation section.
Another query is that it is not clear whether any correction for multiple comparisons was applied. If it was not, it is possible that the significant differences reported may not remain significant. Please add in this detail and justify why corrections were not applied, if they were not.
The reviewer makes a good point here, and between different ANOVA tests for the perceived sensory attributes we do not perform an additional multiple comparison correction. We have however changed that and introduced an extra Bonferroni correction for the sensory attributes, effectively dividing our alpha by 13 in that study.
We comment on this in the Statistical evaluation section, and furthermore, this correction removes the only significant finding we had on sensory attributes, a difference in earthiness. We have changed the text in the Results and Discussion sections accordingly.
Finally, the authors state that between-groups differences were assessed for liking and the 13 different attributes, but it is not clear how/why this can be as I was under the impression that this was a repeated-measures designed. What variables were assessed between-groups? Please clarify.
With “between-group” we refer to groups of apples, not groups of respondents. We changed “between group” to “between apple categories” to clarify this further.
Results – The results of the statistical analyses (e.g. actual f-values and p-values) should be reported or summarised where possible, even for non-significant findings. Currently the authors primarily describe the results in text and it is unclear which analyses are being reported where and how ‘singificant’ and ‘non-significant’ these differences were without any indication of the relevant f-values and p-values.
We choose here to describe to visualize the results in Figures 2 and 3. Furthermore, we added p-values for the significant differences in terms of liking, and the lowest p-values for each sensory attribute to indicate “how insignificant” the differences were in those cases.
We may add a collection of tables indicating all results for the statistical analyses, but we feel that this would consume too much space in the article. If the editor feels that this would be appropriate we will add these tables.
Table 1 legend – P = 0.05 should be P > 0.05?
Thank you, this is corrected!
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The manuscript describes a study where perception of product appearance is linked with product experience and hedonic responses of the experienced products. The results show that appearance of a whole product has a minor impact on liking for tasted samples, but perceived product attributes are not influenced by appearance with some exceptions. The study is carefully carried out and reporting is mostly clear. There are several points, however, that would benefit from clarification.
Major points:
The discussion section contains some parts that are mainly describing results and some that are speculative going beyond what has been studied in the experimental part. Careful reading and re-writing some parts of the discussion would improve the manuscript considerably.
If I have understood correctly, the tasted slices of apples that were linked to the different pictures were just randomly chosen from a pool of eight selected apples. This should result in a disconnection between presented appearance and perception of different experienced sensory attributes as the differences in sensory experience become part of random error. The paper does not justify why this approach was selected and it would be good to know in more detail how the study was run.
What was the reasoning behind selecting this approach?
Did each participant get eight different apple varieties without peel? If yes, did the sensory perception vary according to the varieties consumers tasted?
As the authors acknowledge in the discussion section: consumers may have difficulties in assessing the sensory attributes of samples: why then ask to do that? If the question is how these attributes are linked to liking, why not use correlational analyses across the samples rather than trying to find differences in attribute ratings between different appearance categories?
Why green colour is rated a suboptimal in relation to red colour as this an attribute that depends on the apple variety?
Did respondents receive only one appearance picture or the four pictures depicted in Figure 1?
Minor points:
In the abstract, line 16, “….sub-optimal food product are thought to be…” sounds odd: thought based on what: be more precise
Lines 48-49: vague statement: how may the knowledge help here?
Line 50: “For sustainability to increase” – not sure whether this is a correct expression: please check with language experts
Lines 60-62: specify which of the appearance factors are directly used as definitions of suboptimal products.
Line 72: This cannot be a representative sample, if the reference is to Swedish adult population. I assume you mean a random sample. How were the participants recruited? Combine the information from the beginning of results here (lines 108-112), otherwise you repeat same information.
Lines 102: what between group differences were tested? Did participants taste all eight samples or just some of them?
Line 113: apple categories here probably refer to appearance categories. Use terms in a consistent manner throughout the manuscript.
The whole Result section seems to lack the test values of statistical analyses. Add them and avoid describing difference that are not significant.
Line 153: “foodwaste” should be food waste
Lines 159-163: this is mainly results; move to the result section
Lines 172-175, this is speculation and not relevant for the main research question of this study
Lines 194-202: This is speculation that goes beyond current analyses of your study. It is related to factors you could analyse from your own data: unless you carry out these analyses, these speculations should be omitted.
Lines 221-223: “might be assumed” – this sounds strange: either you can conclude this from your findings or not, be more specific. Check similar expressions throughout the manuscript,
Line 226: not significantly different does not mean significant non-difference.
Line 229-230: this sentence is vague: why increased awareness would be beneficial for sustainability.
Tables and figures should be self-explanatory. Add descriptive explanation on what the samples are and what is the max on the liking scale.
Check the referencing, sometimes the number is missing from the reference or placed in different places in the sentences.
Author Response
Review 2
Dear Reviewer, thanks for comments. We have taken them all into account and revised the manuscript. See below text in red for amendments.
Comments and Suggestions for Authors
The manuscript describes a study where perception of product appearance is linked with product experience and hedonic responses of the experienced products. The results show that appearance of a whole product has a minor impact on liking for tasted samples, but perceived product attributes are not influenced by appearance with some exceptions. The study is carefully carried out and reporting is mostly clear. There are several points, however, that would benefit from clarification.
Major points:
The discussion section contains some parts that are mainly describing results and some that are speculative going beyond what has been studied in the experimental part. Careful reading and re-writing some parts of the discussion would improve the manuscript considerably.
Thank you for the comment. The text in the Discussion chapter has been reviewed and clarified.
If I have understood correctly, the tasted slices of apples that were linked to the different pictures were just randomly chosen from a pool of eight selected apples. This should result in a disconnection between presented appearance and perception of different experienced sensory attributes as the differences in sensory experience become part of random error. The paper does not justify why this approach was selected and it would be good to know in more detail how the study was run.
Thank you for pointing this out. It is now clarified in the text in the Material and method chapter.
What was the reasoning behind selecting this approach?
The study evaluated how visual appearance may influence consumer liking and perception of apples.
Did each participant get eight different apple varieties without peel? If yes, did the sensory perception vary according to the varieties consumers tasted?
This is now clarified in the material and method chapter.
As the authors acknowledge in the discussion section: consumers may have difficulties in assessing the sensory attributes of samples: why then ask to do that? If the question is how these attributes are linked to liking, why not use correlational analyses across the samples rather than trying to find differences in attribute ratings between different appearance categories?
This is absolutely right, however, in sensory science this is of interest when comparing consumers and trained panelists. It may further indicate the importance of the visual appearance for consumers.
Why green colour is rated a suboptimal in relation to red colour as this an attribute that depends on the apple variety?
This is truly relevant. We have therefore added some text in the introduction. Also in the discussion chapter this is discussed.
Did respondents receive only one appearance picture or the four pictures depicted in Figure 1?
One serving with one picture which consisted of the four photos in Figure 1. This is clarified in the text and also a table (Table 1) is added.
Minor points:
Thank you for pointing these out! The points are taken into consideration and changed according to the red text.
In the abstract, line 16, “….sub-optimal food product are thought to be…” sounds odd: thought based on what: be more precise
This is changed and now reads:…. sub-optimal food products is an important…..
Lines 48-49: vague statement: how may the knowledge help here?
“may” is now “is”
Line 50: “For sustainability to increase” – not sure whether this is a correct expression: please check with language experts
The manuscript is reviewed by a language expert.
Lines 60-62: specify which of the appearance factors are directly used as definitions of suboptimal products.
Line 62-63 now reads: These characteristics affect the purchase intentions of consumers, but only if they deviate significantly from the norm and may be perceived as suboptimal.
Line 72: This cannot be a representative sample, if the reference is to Swedish adult population. I assume you mean a random sample. How were the participants recruited? Combine the information from the beginning of results here (lines 108-112), otherwise you repeat same information.
This is now corrected, both line 72 and line 119.
Lines 102: what between group differences were tested? Did participants taste all eight samples or just some of them?
All consumers tested all eight samples. The groups are according liking. The text is clarified, line 105 and statistic chapter. The groups here refer to apple categories, which has been clarified in several places.
Line 113: apple categories here probably refer to appearance categories. Use terms in a consistent manner throughout the manuscript.
The text is clarified
The whole Result section seems to lack the test values of statistical analyses. Add them and avoid describing difference that are not significant.
Line 153: “foodwaste” should be food waste
Corrected
Lines 159-163: this is mainly results; move to the result section
Now more relevant included in the discussion.
Lines 172-175, this is speculation and not relevant for the main research question of this study
Lines 194-202: This is speculation that goes beyond current analyses of your study. It is related to factors you could analyse from your own data: unless you carry out these analyses, these speculations should be omitted.
The text is omitted.
Lines 221-223: “might be assumed” – this sounds strange: either you can conclude this from your findings or not, be more specific. Check similar expressions throughout the manuscript,
“might” is now “can”
Line 226: not significantly different does not mean significant non-difference.
Correct, and we changed to not significantly different.
Line 229-230: this sentence is vague: why increased awareness would be beneficial for sustainability.
Tables and figures should be self-explanatory. Add descriptive explanation on what the samples are and what is the max on the liking scale.
Liking scale is described in the material and method-chapter.
Check the referencing, sometimes the number is missing from the reference or placed in different places in the sentences.
Checked
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Thank you to Normann and co-authors for their revised manuscript. The paper generally reads well and their conclusions appear to support the data and are not overstated. I appreciate that they have attempted to add more information into the methods, but some of the amendments are not very useful and clarity around certain aspects of the methods is still lacking in places and requires revision.
Figure 1 and the description of the 8 apple images was already clear in the previous manuscript (I just could not see the colour-images in my copy). Therefore, the addition of Table 1 seems odd as this appears to be a re-representation of the same information. Perhaps this was recommended by another reviewer? I cannot see any other reviewer reports, so I am not sure why this was added. I suggest having Table 1 OR Figure 1, but not both. I prefer Figure 1.
The part of the procedure that was not clear from the previous manuscript was the description of the 8 ‘random’ apple samples that were tasted, and specifically how they were paired with the 8 different apple images. The authors just say these samples were ‘random’, but this is not adequate. The methods section would benefit from a ‘Procedure’ sub-section, which should detail exactly how the study was ran and what the participants encountered from the beginning to the end of the session (the order that activities were completed and the exact number of trials, any randomisation or counterbalancing procedure etc.). This does not have to be extensive, but the authors need to describe more than the design in order for the reader to actually know what they have done to derive these data.
As part of a Procedure section, the authors should explicitly state the following information:
State whether all participants completed 64 (8 x 8) trials – if this is not the case, please state exactly how many trials they completed, how many required a tasting (if not all of them) and whether all the trails (sensory and liking ratings) were completed in one sitting or separated over a few sessions.
Please state whether the 8 tastes apple samples (of different cultivars grown in Sweden) were pre-selected by the experimenter, not the participants (this is not clear). Please also state whether the same apple sample/cultivar was always paired with the same picture each time, or if this was randomised across participants. If it was randomised across participants (as the text suggests), then how was this randomisation achieved and accounted for in the analyses? Please also state whether the different randomly selected apple cultivar differed in their taste and texture, or whether they were selected to be very similar in taste, texture and appearance. The authors state that “the sensory attributes were not expected to significantly differ between categories”, but it is not clear how their study design was set up to ensure this.
The size of the apple samples tasted was approximately ¼ an apple in size (line 112-113). Please state whether participants were instructed to take a bite of the sample, or to consume it all. If they were no instructions as to how much to consume, please also state this. The participants could have potentially consumed a lot of apple during the testing, which could impact liking ratings.
Please also describe whether the apple samples were served within or without the skin (they should have been served without skin), whether any measures were taken to stop the samples browning, and if so, what, and whether the different apple cultivars were all prepared to look in a uniform way during tasting. If they looked very different, this could have confounded the evaluations of the pictures they were paired with.
Author Response
Thank you for re-review of the manuscript. We have taken all the comments from the reviewers into consideration and made amendments in the manuscript, which now have become more exact and clear. Below, our amendments and comments to Reviewer 1 are written in red text.
Thank you to Normann and co-authors for their revised manuscript. The paper generally reads well and their conclusions appear to support the data and are not overstated. I appreciate that they have attempted to add more information into the methods, but some of the amendments are not very useful and clarity around certain aspects of the methods is still lacking in places and requires revision.
Our aim is to make the text as clear as possible.
Figure 1 and the description of the 8 apple images was already clear in the previous manuscript (I just could not see the colour-images in my copy). Therefore, the addition of Table 1 seems odd as this appears to be a re-representation of the same information. Perhaps this was recommended by another reviewer? I cannot see any other reviewer reports, so I am not sure why this was added. I suggest having Table 1 OR Figure 1, but not both. I prefer Figure 1.
Sometimes the reviewers points and remarks differ. This is the case here, and we suggest the editor to decide what to do.
The part of the procedure that was not clear from the previous manuscript was the description of the 8 ‘random’ apple samples that were tasted, and specifically how they were paired with the 8 different apple images. The authors just say these samples were ‘random’, but this is not adequate. The methods section would benefit from a ‘Procedure’ sub-section, which should detail exactly how the study was ran and what the participants encountered from the beginning to the end of the session (the order that activities were completed and the exact number of trials, any randomisation or counterbalancing procedure etc.). This does not have to be extensive, but the authors need to describe more than the design in order for the reader to actually know what they have done to derive these data.
Thank you for pointing this out. We realise that the text needs some clarification and therefore the section has been partly rewritten in order to avoid misunderstandings.
As part of a Procedure section, the authors should explicitly state the following information:
State whether all participants completed 64 (8 x 8) trials – if this is not the case, please state exactly how many trials they completed, how many required a tasting (if not all of them) and whether all the trails (sensory and liking ratings) were completed in one sitting or separated over a few sessions.
See above, it is now stated that each consumer participated in eight trials, see line 72-74.
Please state whether the 8 tastes apple samples (of different cultivars grown in Sweden) were pre-selected by the experimenter, not the participants (this is not clear). Please also state whether the same apple sample/cultivar was always paired with the same picture each time, or if this was randomised across participants. If it was randomised across participants (as the text suggests), then how was this randomisation achieved and accounted for in the analyses? Please also state whether the different randomly selected apple cultivar differed in their taste and texture, or whether they were selected to be very similar in taste, texture and appearance. The authors state that “the sensory attributes were not expected to significantly differ between categories”, but it is not clear how their study design was set up to ensure this.
Please, see lines 99-102 where this is stated.
The size of the apple samples tasted was approximately ¼ an apple in size (line 112-113). Please state whether participants were instructed to take a bite of the sample, or to consume it all. If they were no instructions as to how much to consume, please also state this. The participants could have potentially consumed a lot of apple during the testing, which could impact liking ratings.
The text is clarified to include this information.
Please also describe whether the apple samples were served within or without the skin (they should have been served without skin), whether any measures were taken to stop the samples browning, and if so, what, and whether the different apple cultivars were all prepared to look in a uniform way during tasting. If they looked very different, this could have confounded the evaluations of the pictures they were paired with.
Please, see lines 99-102 where this is stated.
Author Response File: Author Response.pdf
Reviewer 2 Report
Most of my concerns have been responded to. Below are two suggestions that would improve the paper further:
Thank you for adding the Table 1. It would be more informative if the columns contained in the cells the information that is now in the Table heading (green - red for colour).
Referring to differences that are not statistically significant in the abstract and later in the discussion should be taken out. These can be just random findings.
Author Response
Thank you for re-review of the manuscript. We have taken all the comments from the reviewers into consideration and made amendments in the manuscript, which now have become more exact and clear. Below, our amendments and comments to Reviewer 2 are written in red text. Our aim is to make the text as clear as possible.
Comments and Suggestions for Authors
Most of my concerns have been responded to. Below are two suggestions that would improve the paper further:
Thank you for adding the Table 1. It would be more informative if the columns contained in the cells the information that is now in the Table heading (green - red for colour).
Thank you for the suggestion! This is now added to the table.
Referring to differences that are not statistically significant in the abstract and later in the discussion should be taken out. These can be just random findings.
We saw a very clear trend in the data, which was close to significant. We have clarified the text that this is a clear trend, however not statistically significant. The discussion is also about using consumers to describe perception compared to using a trained sensory panel – we see this as an important part of the discussion.
Author Response File: Author Response.pdf
