All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
Thank you for the revised version, having added sufficient detail with regard to the recruitment process and an explanation as to the inability to calculate response rates, you paper is now accepted for publication
Thank you for your submission. Please address the issues raised by reviewer 2 with regard to your reporting of response rate (this should be a proportion of those invited) and the provision of additional detail on participant selection so as to allow full understanding of the recruitment process.
See below
See below
See below
I am generally happy with the responses to this second revision. In particular, I am content with the sample being described as a purposive sample to avoid any confusion. It's a sensitive topic at the moment because there has recently been a 'prevalence' study conducted in Africa (gaming area) which has been pulled to pieces online (by Nick Brown, the research integrity blogger who may be on the warpath against the gaming area). Very important that any paper in this area is tempered and balanced in its conclusions.
I actually agree about the limitations relating to probability sample. I've recently seen some post-weights in prevalence studies which are enormous and questions could definitely be raised about whether these cases (young males in particular) are representative of the population. Using some quota sampling or stratification can be useful. This is a topic where I've had lengthy chats with Rob Williams who has just run the Canadian prevalence study. I think that researchers run into problems if you post an add (e.g., on Reddit or somewhere else online), get people to self-select/ opt-in and then try to make it sound like a population sample if the basic demographics look like the population (or are weighted to be the way). You never get past the response bias (respondent 'I do this because this is about people like me!'). Bob Ladouceur's paper on volunteer sample biases might be worth citing (in 1990s). You get over-estimates..... which I think is what you've found too for gambling vs. the BPS for some forms of gambling?
The best strategy may be to use probability method to identify your panel in advance (Rob's view). The HILDA sample in Aust is like this. Do all you can at the outset to get a panel via random methods and then try to fill in the gaps until you get a final cohort that requires as little weighting as possible. This means that you haven't just got people who have opted in or self-selected to take part because it's about gambling or LBs or gaming. In effect, everyone in the population has a chance to take part at the outset (probability method), but you then use other methods to make it representative.
It would be in line with normal reporting also to insert a reference for APA in line 353 although it has been sited eralier.
Regarding response rate I still disagree with the author. Response rate is not the number who participated. Response rate is the proportion of those who who actually participated to the number of those who were invited (or who could participarte). I cannot see that his number is provided. This number is important in all epidemiologoical research as it says something about the representativity of the sample. In the abstract it is stated that 1081 participated.
In the methods it is stated that an initial sample of 1201 was collected. Further you state that sample was recruited for us by Prolific Academic. How did this take place? How did Prolific Academic recruit them - from where? How many did Prolific Academic ask in order to obtain the sample? Thats important issues to comment on and to address properly.
N/A
N/A
Thank you for the revised submission. While the work has addressed some of the concerns raised by the reviewers, there remain a number of issues that need to be resolved. In preparing your response to reviewers it would be appreciated if you could signpost the changes in your manuscript so that the reader can quickly assess the amended content.
Please ensure that you address each of the points raised by the reviewers either by amending your submission, or presenting your argument as to why no change was made.
Better
It's correlational.
A few comments below.
2nd review
I thought that this version of the paper was much improved and provides a more balanced and logical coverage of the material. However, there are still a few points which raise some concerns for me.
Abstract: It IS a convenience or panel sample. No way that this sample was obtained by probability methods. All the references in the world re. value of stratified samples can be cited, but it still cannot be read as a prevalence study. I still think this is misleading and not good scientific practice because journalists won’t know the difference.
Table 6: Don’t like the word ‘influence’. Association better.
In the Discussion there is a reference to 18.5% of gaming behaviours being on par with other forms of gambling, but wouldn’t the fairer comparison be with the overall participation rate for gambling? I’d cite the most recent BPS and this will be much higher than 18.5% for the annual participation rate for any form of gambling?
The paragraph that starts on Line 512 I’d start off with the more plausible explanation in the abstract: PGs who play video games probably like loot boxes rather than loot boxes really having any influence on whether people take up gambling. The evidence barely shows this. There is really only the Molde et al (2020) that shows one cross-lagged effect and it’s small: .15 between problem gaming at Wave 1 and PGSI at Wave 2.
566-569: I didn’t find this all that convincing. A lot of male PGs will gamble online in the UK, and they will ALSO gamble on land-based activities. Yes, they have other land-based activities, but they spend a lot of time online and this would increase the likelihood of online gaming and loot boxes.
586 paragraph. It IS a convenience sample.
A broader Q: are LBs really a major problem in terms of harm? Is $20-30 per month (as found in the other papers) really a lot? It is enough to cause harm?
Luke Clark is singular? Their? Luke is not identifying as non-binary as far as I know.
Just one comment on the rebuttal. Just because others are saying the same thing- doesn’t strengthen the argument. An example: I have a lot of doubts about adolescent problem gambling prevalence figures (too high), but lots of us have got these results. It’s a reliable finding, but is it valid? The measures might be consistently misinterpreted by young people.
I asked the authors to provide a definition of both problem and disordered gaming. Still, I cannot find that the latter (disordered gambling) has been done.
Also, the rebuttal latter is not acceptable. Either the authors should link (e.g. by line no) the answers to specific places in the manuscript or specifically in the rebuttal letter write how they have changed the ms. Just writing in the letter that this and this have been augmentet is quite arrogant, and make the job as reviewer much more time-consuming than neccessary.
Please use correct reference for the DSM-5 (line 353).
I asked the authors to say something about the response rate. They refuse do to so, and just repeat the number of participants. N and response rate is not the same thing. I also searched the whole ms for the expression "response rate" and it did not appear.
I probed the authors about the representativness of the sample. The authors claim that the sample was stratified, but use the term quata sampling instead - withouty defining the difference between stratified and quota sampling.
Also, even if a sample is stratified or based on quotas, different response rates in different quotas may make it non-representative. This was my point, unfortunately the authors have not adhered to this!
I asked also for a rationale for the independent variables. The authors just reply that age and gender now have been included. But thats not the same as providing a rationale.
I asked the authors about third variables (e.g. impulsivity) - but could not find anything explicitly about this in the revised ms.
Gaming-related behaviors explained only approximately 5% of the variance in problem gambling - instead of being humble the authors state: In fact, we would argue that it is instead the case that the relationship between gaming-related behaviours and disordered gaming is surprisingly large, and bears further study. So when did 5% become surprisingly large?
When you make a rebuttal letter you need to be much more specific - incorporate all changes in the letter - it is not sufficient to say that the ms has been augmented.
Thank you for your interesting submission. All three reviewers have indicated that the study presents interesting findings, however they have all provided comments and suggestions that must be addressed before we can consider your paper for publication.
.
.
.
PeerJreview-Lootboxesandgambling
gambling and the purchase and/or use of loot boxes in video games. This has led to speculation about whether loot boxes are a new form of risky gambling-like activity that is appealing to problem gamblers. The paper also tentatively raises the question as to whether loot boxes might be a softer gateway product to engagement into gambling which, in turn, can lead to problem gambling.
I was generally impressed with the sample size and choice of measures in this paper. The data and results are publishable, but there were many aspects of the paper which need to be revised to make it a stronger paper.
The first problem is that the study is not really a prevalence study. Stratification might make the sample superficially resemble the age and gender, ethnicity (probably mostly British vs. the rest), but Profilic samples and any survey of this nature will be heavily skewed towards people who are more highly educated and who like to s This paper presents the findings of an online panel survey that investigates the association between various forms of gambling behaviour, gaming and particular aspects of gaming, e.g., the use of loot boxes. The sample is drawn from Prolific and uses a stratified sampling method to attempt to make the sample reflect some of the demographic characteristics of the general adult population in the UK. A sample of over 1000 people was obtained and the study included a number of respectable and standardised measures using the PGSI and Lemmen’s measure of gaming disorder.
The main finding of the paper (which replicates other similar studies by the author) is that there is a statistically significant relationship between it at home and do surveys. It will also tend to attract people who have an interest in the topic. Thus, any estimates will be far higher than what might obtain in the general population. I regard with significant suspicion any study that reports a prevalence of problematic gaming above 2-3%. The same holds for any figure above 1-2% for problem gambling. You just don’t get this sort of figure in probability sampling. Lemmen’s measure also tends to score higher than the Petry measure. All references to prevalence should be strongly qualified or removed. It’s really a convenience sample.
A second problem is that the study does not really engage with selection and exposure effects. I find the ‘gate-way’ argument to be long stretch and found a lot of the argument in this paper to be too causative. Gambling and gaming are largely different classes of activity. Loot boxes might look like gambling in the context of video games, but they have nothing to do with gambling as we regularly know it. One does not gamble on loot boxes when one visits a high street betting outlet, visit a casino or when one gamblers on Bet365. A stronger argument is that this is merely a selection effect. The sorts of people who like to gamble ALSO like to play video games and, if you are really into video games, you tend to buy loot boxes. In other words, loot boxes are just another way to extract money from people who also like to gamble a bit more (probably young men). The two run in parallel. I’d take out all the linkage stuff which is unfounded and definitely not talk about ‘gate-ways’.
A bit more could be made of technological convergence- that people who go online to gamble or use their devices a lot also have a chance to play games as well.
A third problem is that the paper does not control for gender. I think the regression model needs to do this. Males will play more VGs, open more loot boxes and also gamble a lot more.
The writing needs lots of editing to tighten up the argument.
Abstract: This reports involvement almost like a prevalence study. Not what this sample should be used for.
You should not be using phrases like ‘gaming must urgently be investigated’. Far too rhetorical. This is not a call to action or activist paper. Needs to stay neutral and academic in style.
Line 37: I felt that this was over-stated too. Only the author is advancing this argument. The link and casual link is tenuous. Don’t talk about ‘concerns’ and build up the rhetoric to sell the paper. OK in a grant, but shouldn’t be in a paper. Line 38 is not a sentence.
Line 44 This paragraph needs to be reworked. Syntax seems odd.
Line 48 This needs to be revised ‘gate-way’ etc.
I found the listing off all the other activities in small paragraphs a bit choppy. Could this be put into a series of (a), (b), (c) indented paragraphs. Otherwise, feels disjointed.
-Very few people actually pay money to get extra credit on SCGs. The prevalence of ‘social casino spending’ is NOT high. It might be 1%.
-The logic in the line 107-115 section seems out to me. VGs and problem are parallel activities. I don’t think there’s much of a link of the nature described here.
Line 128 ‘to begin investigating’. Hasn’t the author published on this before?
Line 131: No it’s not a prevalence study
Line 331 sounds too much like prevalence; linked is too causal.
Iine 355 Not sure about the meaning of clinically significant effects.
Next paragraph: a lot of the differences could be due to sampling.
The conclusion starts to grapple with the causality issue, but this should have been said earlier (line 420 onwards)
Overall, a potentially useful set of data, but needs a bit more analysis of gender; a lot of editing and bit of restructuring of the argument; and more tempered conclusions about the causal relationship between activities.
The present paper presents data concerning a survey investigating how gambling problems and disordered gaming are related to opening loot boxes, esports betting, real-money video gaming, token wagering, and social casino play. A representative sample of 1081 adults from the UK aged 18+ was recruited for the study.
Major issues:
Line 106: By which standard can it be said that an eta-squared of .04 is clinically significant. Please back this statement with clinically relevant empirical findings.
In the introduction the author uses the terms problem gambling and disordered gaming. These concepts should be defined.
Line 143. Although an initial sample of 1201 was recruited – what was the response rate – e.g. how many was approached – how many could not be reached and how many refused to participate?
Please place the description of the instruments before the description of sample characteristics, as the latter, e.g. gambling category, is based on the former.
Minor issues:
Line 37. The sentence starting with “gambling like transactions” seems to be missing something.
Line 140: Please explain what you mean by “self-reported remotely”
Line 161: Please don’t start a sentence with a number
Please also check that spaces between words are as they should be (e.g error in line 372, 389) and that the same font size is used allover.
In terms of the PGSI – please state that the scoring procedure deviates from what is most common – and please state what defined disordered gambling. Also, provide information about the Cronbach alpha for the scale.
Line 249. The acronym APA has not been introduced to the reader. Please state that the Internet Gaming Disorder Scale (use capital letters) is based in the diagnostic criteria for Internet Gaming Disorder found in the 5th edition of the Diagnostic and Statistical Manual for Mental Disorders. It seems (table 2) that you used continuous scores based on the IGDS in some of the analyses. Please provide information about this. Also provide information about the Kuder-Richardson-20 (equivalent to alpha for scales with dichotomized response alternatives).
Although the original sample was representative it might be different response rates in different demographic groups (e.g. typical lower response rates among males and young people). Did you adjust for the discrepancy between sample and the UK population in terms of important demographic variables?
Table 1: Explain the acronyms LLCI and ULCI.
What was the rationale for selecting just some potential variables as independent in the regression analyses? A better/or detailed justification is in order. I think you should consider including age and gender as independent variables.
Regarding the regression analyses: Did you check that there were no violations of the assumption of normality, linearity, multicollinearity and homoscedasticy?
Regarding the discussion: There does not have to be any causal link between study variables. The associations may for example reflect a common underlying third variable like age, impulsiveness, interest in gambling and gaming etc.
It is also possible that the common method bias may play a role as self-report data only, with a cross-sectional design, were collected – causing inflation in the estimated associations between study variables.
Please also comment on the fact that gaming related activities explained only 5.6% of the variance in problem gambling whereas they explained 21.7% of the variance in problem gaming. Hence, could it be argued that the convergence between gaming and gambling is not as large as suggested by the author? Please discuss this issue further.
Please also suggest some directions for future research (e.g. longitudinal studies).
English is not my first language, but to me, the manuscript appears to be written clearly and to the point. It succeeds at communicating the main research questions, findings and discussion, without using unnecessary jargon or overly complicated statistical analyses.
The tables are adequately labelled and explained, and raw data, as well as complete item descriptions, are supplied via the OSF repository.
The study appears adequately designed to address the research questions outlined in the “Summary” section of the introduction.
The investigation is conducted and presented using fairly standard statistical approaches, and I found the reporting adequate.
The survey items are included via the OSF repository, enabling anyone interested to replicate.
I am aware that novelty is not a requirement for PeerJ, but would still like to commend the author for collecting this data. The results should be useful for any party interested in the grey area between traditional gambling and video games. I do, however, have some specific issues that the author may want to consider.
1. Representativity. Although it is not claimed that the sample is representative for the UK, steps taken to ensure representativity in terms of ethnicity, sex and age are mentioned. I am unfamiliar with the details of how Prolific Academic works, but am under the impression that it is a service where people self-select to partake in an online panel, where they can be invited to surveys and be paid for their time. I would like to see more details but suspect that the resulting sample cannot be regarded as representative of the UK 18+ population due to the nature of how PA service works (i.e., not all of UK are members of this panel and those who are might differ from those who are not). Also, information about response rates would be useful in this regard, if at all available.
2. Incentives. Were participants paid, and if so, how much
3. Esports betting. I think a case could be made for categorizing this activity as a form of gambling rather than a “gambling-like” or “gambling-related” practice in video games (as it is described in the introduction). I think it would make more sense to group this with the 11 traditional forms of gambling – since this activity is gambling and usually offered by gambling companies rather than the video game industry. I would not, however, see this as a crucial change and would leave it to the author to decide which approach is most suitable. I do, however, believe that some clarification needs to be made. In the introduction, it is referred to as “a variety of gambling-like and gambling-related practices … in video games” (line 40), and then “gambling-like video game practices” (line 130). In the measurements section, however, it is referred to as “gaming-related forms of gambling and gambling-like behaviour” (line 205). I believe e-sports betting fits the latter description, but less so the first two. Regardless, please be consistent in how these concepts are described and referred to.
4. Some of the introduced concepts are probably unfamiliar to a large proportion of the respondents. In particular, I believe many will have little intuitive understanding of what “Real money gaming” and “Token wagering” entails. For example, although I have a keen interest in video games, both academically and as a hobby, I am not entirely sure if I have taken part in token wagering even after reading the explanation provided in the survey (OSF repository). I believe some discussion around how this might have influenced the findings would be warranted.
1. I would like to thank the author for conducting this exciting piece of research and believe it is a valuable contribution to the field. However, after checking the complete list of items provided in supplemental materials, I think there are several additional and relevant points that should be included in this manuscript. Including:
-Was there a relationship between problem gambling and disordered gaming?
-What was the proportion of purchasing loot boxes among people playing games that included these products?
-Data from the “what came first” items would be very interesting and relevant information.
-More information about the video game habits of responders could be provided
2. The proportion of respondents with problem gambling or disordered gaming appears to be quite high compared to other studies (I believe the UK Gambling Commission reported 0.7% of problem gamblers, compared to 2.2% in the present study). I think these numbers should be discussed. Could a lack of representativity be a part of the explanation? (see the previous comment under “Validity of the findings”).
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.