The changing face of desktop video game monetisation: An exploration of exposure to loot boxes, pay to win, and cosmetic microtransactions in the most-played Steam games of 2010-2019

David Zendle; Rachel Meyer; Nick Ballou

doi:10.1371/journal.pone.0232780

. 2020 May 7;15(5):e0232780. doi: 10.1371/journal.pone.0232780

The changing face of desktop video game monetisation: An exploration of exposure to loot boxes, pay to win, and cosmetic microtransactions in the most-played Steam games of 2010-2019

David Zendle ^1,^*, Rachel Meyer ¹, Nick Ballou ²

Editor: José C Perales³

PMCID: PMC7205278 PMID: 32379808

Abstract

It is now common practice for video game companies to not just sell copies of games themselves, but to also sell in-game bonuses or items for a small real-world fee. These purchases may be purely aesthetic (cosmetic microtransactions) or confer in-game advantages (pay to win microtransactions), and may also contain these items as randomised contents of uncertain value (loot boxes). The growth of microtransactions has attracted substantial interest from both gamers, academics, and policymakers. However, it is not clear either how frequently exposed players are to these features in desktop games, or when any growth in exposure occurred. In order to address this, we analysed the play history of the 463 most-played Steam desktop games from 2010 to 2019. Results of exploratory joinpoint analyses suggested that cosmetic microtransactions and loot boxes experienced rapid growth during 2012–2014, leading to high levels of exposure by April 2019: 71.2% of the sample played games with loot boxes at this point, and 85.89% played games with cosmetic microtransactions. By contrast, pay to win microtransactions did not appear to experience similar growth in desktop games during the period, rising gradually to an exposure rate of 17.3% by November 2015, at which point growth decelerated significantly (p<0.001) to the point where it was not significantly different from zero (p = 0.32).

Introduction

The way that the video game industry makes money has undergone important changes in recent decades. In the 1990s and early 2000s, industry profits were largely based around the sale of copies of games [1]. These copies might take the form of cartridges, discs, or even digital downloads. Under this model individuals were handing over money in return for the either the ownership of a complete product, or the license to play that product for a potentially unlimited period of time [2]. Similarly, ownership of a product might occur via a subscription-based model: It has been common for decades for players of online role-playing games to pay a flat monthly charge for access to a game.

However, at some point in the early 2000s, monetisation in video games underwent a significant shift. As well as selling games as complete products, publishers also began offering gamers the ability to purchase additional items, bonuses or services within the game itself for a real-money fee, known as a ‘microtransaction’ [3].

Cosmetic microtransactions

As noted in [4], many microtransactions allow players to purchase decorations and alternative costumes that “offer no in-game advantage and are purely aesthetic“. In the context of this paper, we refer to any situation in which spending additional money leads to an aesthetic change within a game but no in-game advantage as a ‘cosmetic microtransactions’.

The cosmetic microtransactions that may be made in video games are varied. For example, in the multiplayer battle royale game Fortnite, players can spend real-world money to buy in-game ‘emotes’ that allow them to express ideas and feelings via the movements of their in-game avatar. In the vehicular soccer game Rocket League, players can pay to purchase new ‘goal explosions’ that allow them to celebrate in-game victory with unique visual effects. And in the third-person shooting game Anthem, players can buy new armour pieces for their in-game mechanical suits of armour. These pieces do not confer any in-game boosts or advantages in terms of fighting: They simply look different.

Pay to win microtransactions

However, as noted in [5], not all microtransactions in games are purely cosmetic in nature. Players of many modern video games are also given the option to purchase virtual items and bonuses that increase their chances of in-game success. In this paper, we define any situation in which players are able to exchange real-world money for something that increases their chance of in-game success as a ‘pay to win’ microtransaction.

Some ‘pay to win’ microtransactions do not have any effect on the aesthetic of a game. For example, players of the multiplayer mode in The Last of Us can pay real-world money for advantages such as the ability to sneak up on other players silently via an ‘Agility perk’. This in-game advantage does not change how the game itself looks: It merely alters how the game is played.

However, other “pay to win” microtransactions also change how a game looks. For example, players of the game Awesomenauts can spend real-world money to purchase additional in-game characters. These new characters can convey an in-game advantage. However, they also have unique and special looks, and therefore have cosmetic value as well. Within the context of this paper, if a microtransaction changes both how a game looks and confers an in-game advantage, we would categorise that microtransaction as ‘pay to win’.

It is import to note that some games separately offer both cosmetic microtransactions and pay to win microtransactions. An example of this is Assassin’s Creed: Odyssey. In this game, players may pay real-world money to purchase a boost that enables them to level up more quickly, but does not change how the game itself looks. This is a pay to win microtransaction. Conversely, they may spend real-world money to purchase a ‘skin’ for their in-game mount that changes how it looks, but does not affect gameplay–this is an entirely cosmetic microtransaction.

Pay to win microtransactions are thought to have originated with online multiplayer games such as MapleStory in the early 2000s [6], and have garnered controversy amongst both gamers and academics alike. Criticisms of pay to win microtransactions are wide-ranging. Some academics provide ethical critiques of how they may change “the game from a competition where the best player wins to … who wants to and can pay the most” [7]; others posit a belief that this model makes games unfair for less affluent players [8]. In [9], researchers suggest that they may encourage the entrapment of players. They posit that games such as Candy Crush may set up situations in which in-game goals are almost attained (‘near misses’) in order to encourage pay to win purchasing; and that this strategy may lead to continued play and spending “to the similar extent of wins as demonstrated in the gambling literature”.

Controversies over pay to win have led some game developers to explicitly reject these microtransactions as an element of their design philosophies [10]. Furthermore, despite the popularity of games with pay to win elements, many individuals have publicly voiced their displeasure with their incorporation in the games that they play [11].

Loot boxes

As described above, microtransactions can lead to both aesthetic differences, and gameplay advantages. However, when making a purchase, players are not always aware what advantage or difference they are buying due to a monetisation strategy known as loot boxes. A definition of loot boxes is given in [12] as follows:

Loot boxes are items in video games that may be bought for real-world money, but which provide players with a randomized reward of uncertain value

This may be considered a restrictive definition of loot boxes: For example, a sceptic might suggest that in-game items rewarded purely through gameplay be considered loot boxes. We would defend the above definition on the basis that it explicitly invokes the potential for monetisation, which sits at the heart of many issues regarding loot boxes. Furthermore, it was used in oral testimony to a recent UK Parliamentary Select Committee, and was subsequently used by this legislative subcommittee in their official report regarding the potential for harm present in loot boxes [13, 14]. It therefore provides a widely-used and useful definition of loot boxes. This definition is used throughout this paper.

Loot boxes take diverse forms. Some may be considered pay to win: For example, players of the fighting game Marvel: Contest of Champions may pay real-world money to open sealed in-game crystals that contain characters from Marvel franchises. Owning powerful and rare characters can help the player win in-game fights. However, when a player hands over their money to open a crystal, they have no way of knowing whether the character that crystal contains is a rare and powerful one, or a weak and common one.

Others loot boxes may be considered purely cosmetic microtransactions. For example, players of Counter-Strike: Global Offensive may spend real world money to unlock sealed ‘weapon cases’. Each case contains a novel aesthetic for an in-game gun or knife. However, when paying to open a weapon case, players do not know which cosmetic upgrade they are paying for.

Loot boxes are thought to be extraordinarily lucrative for the video games industry, with one source estimating that they may have generated as much as $30 billion in revenue in 2018 alone [15]. However, there are distinct concerns about this monetisation strategy. As noted in [16], loot boxes share distinct similarities with gambling. This has led to concerns that engaging with loot boxes may lead to increases in gambling amongst gamers [17]. Evidence for this causal mechanism is unclear. Spending on loot boxes has been repeatedly linked to problem gambling. However, it is uncertain whether this is because loot boxes cause problem gambling, or whether it is because individuals with pre-existing gambling problems spend more money on loot boxes [18–20].

The present research

It is widely acknowledged that both pay to win microtransactions, cosmetic microtransactions, and loot boxes have become more common in recent years. This has been accompanied by substantial interest.

However, how these features are changing over time is unclear. For example, some news reports have recently suggested that loot boxes are currently becoming more widespread [21], whilst others report that loot boxes are currently in decline [22]. Still more imply that the prevalence of specific in-game features may render them relatively unimportant: A recent statement from one industry representative characterises loot boxes as “a particular form of randomised in-game purchase which feature[s] in a minority of games” [23]. However, to the best of our knowledge no piece of academic research has investigated changes in exposure to either loot boxes, pay to win microtransactions, or cosmetic microtransactions.

This piece of research therefore sets out to explore the changing rate of exposure to loot boxes, pay to win microtransactions, and cosmetic microtransactions by analysing historical data on how many individuals play games with these features each day.

The Steam platform is often considered to be the dominant way for desktop video games to be both sold and delivered [24, 25]. In this piece of research, we create a dataset of the number of players of each of the most-played Steam games. This dataset records the peak number of simultaneous players for each game on each day from the 22^nd March 2010 to the 22^nd April 2019. We then code each of these games for the presence of loot boxes, pay to win microtransactions, and pay to win microtransactions. We then explore how these features change within the sample over time via a joinpoint analysis.

Method

Ethics

This research consisted of an analysis of SteamDB data. SteamDB is a publicly available database that lists the daily number of players of a variety of games. For example, the daily number of players of Counter-Strike can be viewed by browsing to https://steamdb.info/app/730/.

This study solely makes use of SteamDB data. Due to the publicly available and naturally anonymous nature of the aggregate data used in this study, ethical approval was not applied for when conducting this study. Upon submission of this manuscript, a formal waiver from the lead author’s host institution was requested by journal staff. Said waiver was applied for and granted by the ethics officer for the lead author’s department on the basis that this project uses a publicly available database of information that is not personally identifiable.

SteamDB is a browsable database of aggregate play data from a variety of desktop games. The design and conduct of this research did not violate the terms and conditions of SteamDB.

Design

A list was made of the all-time most-played desktop games on the Steam platform. This was operationalised as any game that had achieved over 10,000 simultaneous players. This led to the creation of a list of 474 games that fit this criterion on the 22^nd April 2019 via reference to the SteamDB website [26], which keeps a record of the peak number of simultaneous players for each game on the Steam platform.

The complete play history of each of these games was then extracted in turn from SteamDB. Inspection of these records revealed that a daily log of peak simultaneous players was kept for each game by the Steam platform from 22^nd March 2010.

This process was achieved by first navigating a browser to the SteamDB website, where a list of games ordered by all-time peak simultaneous players is displayed. All entries with 10,000 or more simultaneous players were noted down by a researcher. The researcher then used a browser to navigate to the SteamDB page for each of the 474 games. A link on each of these pages allowed the direct download of a.csv file containing the complete history of the number of players of that game.

Measures

The following three variables were then measured for each of these games:

The presence of loot boxes,
The presence of pay to win microtransactions
The presence of cosmetic-only microtransactions.

The presence of loot boxes

Using the definition of loot boxes given earlier, coders were instructed to record that a game tested positive for the presence of loot boxes if it contained in-game items that could be bought for real-world money but which contained randomized rewards. An example of a game that would test positive for loot boxes is NBA2K18, a basketball game in which gamers can pay real-world money to purchase ‘player packs’ that contain a randomised assortment of new basketballers for their team. An example of a game that would test negative for loot boxes is The Elder Scrolls: Oblivion. Players of this game could pay real-world money to purchase new in-game content (for example, armour for their horses). However, when handing over their money they always knew what they would get in return.

The presence of pay to win microtransactions

Using the definition of pay to win microtransactions given earlier, coders were instructed to record that a game tested positive for the presence of pay to win microtransactions if players could pay real-world money to in any way increase their chances of in-game success. An example of a game that would test positive for pay to win microtransactions is Grand Theft Auto V, in which players may pay real-world money for in-game currency, that can be used to purchase powerful new weapons. A game that would test negative for pay to win microtransactions is the team-based strategy game DOTA 2. In this game, players can pay real-world money to unlock new aesthetics for their in-game characters: However, spending money can never confer an in-game advantage.

The presence of cosmetic microtransactions

Using the definition of cosmetic microtransactions given earlier, coders were instructed to record that a game tested positive for the presence of cosmetic microtransactions if players could pay real-world money for things that offered no in-game advantage and purely led to an aesthetic change. An example of a game that would screen positive for cosmetic microtransactions is Rocket League, in which a variety of decals, goal explosions, and other aesthetic effects may be bought for real-world money. However, none of these purchases are able to change how the game is played. A game that would test negative for cosmetic microtransactions is the digital collectible card game Artifact, in which players can pay real-world money to purchase new cards, all of which have some theoretical gameplay value.

It should be noted that some games could be coded as containing both cosmetic and pay to win microtransactions: A good example of this is Assassin’s Creed: Odyssey, as detailed in the literature review. Similarly, should a game contain loot boxes, it would also be coded as containing ‘pay to win’ or ‘cosmetic’ microtransactions on the basis of whether those loot boxes contained cosmetic rewards or pay to win rewards. NBA2K18, for example, whose randomised ‘player packs’ contain new basketballers that give gamers an in-game advantage, would be coded as containing both pay to win and loot boxes.

A final note should be made regarding the time at which the presence or absence of these features was recorded. The presence or absence of all of the above features was coded on the basis of those features currently appearing in-game at the time of analysis. It was deemed infeasible to consistently determine whether games had added or removed these features at any point during the period under study (2012–2019). As covered in our discussion, some games may have inserted or removed loot boxes, pay to win microtransactions, or cosmetic microtransactions during the period. This is considered in our discussion.

The presence of each of the features outlined above were measured by having two researchers separately code each game for their presence or absence. A single illustrative example of Counter-Strike was provided as an exemplar at the beginning of the coding process. Following this, coders explored the presence or absence of relevant in-game features primarily through a combination of reading the game developer’s documentation and descriptions, and searching for other information regarding the games online such as in forum posts. If this was insufficient, researchers engaged in watching videos of others playing the games and, as a last resort, playing the games in question themselves.

An initial round of coding resulted in near-perfect agreement between coders when it came to the presence of loot boxes (97%, Cohen’s Kappa = 0.90). However, there was only substantial agreement when it came to the presence of pay to win (85.5%, Cohen’s Kappa = 0.66) and cosmetic microtransactions (84.6%, Cohen’s Kappa = 0.68).

Disagreement between human coders is an extremely common feature of any reliability analysis. Indeed, as noted in a standard textbook on the topic, for many researchers, a raw agreement rate of 80–90% is taken as sufficient in a variety of contexts [27].However, given the importance of a reliable coding scheme to our analyses, we resolved to be more stringent. Cohen’s Kappa measures the degree of agreement between coders when chance is taken into account, and is the most commonly used way to measure inter-coder reliability. A Kappa statistic of greater than or equal to 0.81 is categorised according to multiple common benchmarking schemes as representing ‘near perfect agreement’ [28]. As in previous work on similar topics (e.g. [12]), it was determined that after achieving a minimum acceptable Kappa level, any remaining disagreements between coders would then be resolved through dialogical intersubjectivity to yield a dataset whose accuracy we were confident in.

Disagreements in coding were first discussed before re-coding the data. From these discussions, it emerged that disagreements in coding may have been due to a lack of clarity about whether downloadable content (DLC) such as expansion packs should be classified as either pay to win or cosmetic microtransactions. This is a subtle point. The simulation game Farming Simulator 15, for example, has 4 DLC releases that contain new branded machinery that may minorly improve a player’s farming capabilities. Should this be classed as a game with pay to win microtransactions?

In order to resolve this, it was agreed that cosmetic and pay to win microtransactions would be classified as in-game items and rewards that are purchasable with real-world money but do not add substantial additional game content. This was undertaken in order to distinguish as best as possible between the addition of small amounts of additional content via microtransaction, and the offer to purchase substantial video game expansion packs such as in Skyrim. For example, the Echoes of Auriga Pack in Endless Legends may give the player in-game skins such as the Drum of Gios. However, it also comes with a substantial additional content in the form of a new soundtrack, and thus was not coded as a cosmetic microtransaction.

Every game in the dataset was then recoded separately by both coders using this new definition. This round of coding led to near-perfect agreement for both pay to win (96.5%, Cohen’s Kappa = 0.91) and cosmetic microtransactions (96.3%, Cohen’s Kappa = 0.92). Eleven games remained uncoded at this point. These were either demos, test servers, or other non-game products (e.g. an SDK).

Both coders then met and discussed the remaining games on which their codes conflicted. The resolution of these cases via dialogic intersubjectivity led to perfect agreement, and a final dataset of games annotated with the presence of both loot boxes, pay to win features, and cosmetic microtransactions.

Overall, 463 games were included in the final dataset after removing the eleven that could not be categorised. There were 75 games with loot boxes, 388 games without them, and 11 games that could not be categorised. There were 135 games with pay to win microtransactions, 328 games without them, and 11 games that could not be categorised. There were 203 games with cosmetic microtransactions, 260 games without them, and 11 games that could not be categorised.

The presence of both multiplayer and co-operative features were additionally measured for each of these games. Analysis of these features is not presented here.

Changes in exposure to loot boxes, pay to win, and cosmetic microtransactions was measured by first recording the number of players of each game under test for each of the 3,319 measured days from 22^nd March 2010 to 22^nd April 2019. This was accomplished by extracting the complete history for each game from the SteamDB website. Any missing days were filled in via linear interpolation. The total number of players was summed for each day. The number of players of games with each specific feature on each of these days was then calculated. This figure was then divided by the total number of players overall for that day, and multiplied by 100 to yield a percentage measure of exposure.

Statistical analysis

Changing trends in video game features were explored using joinpoint regression. Joinpoint regression is a technique for procedurally fitting a segmented regression model to trend data in order to identify points in a dataset at which a trend changes [29]. It begins by fitting a linear model to the dataset under test, and then iteratively tests whether the segmentation of this model via one or several ‘joinpoints’ leads to an improvement in overall fit. Joinpoint regression is suitable for the analysis of time series data, and commonly used to analyse change in trends over time. It is most commonly used in the analysis of changes in cancer rates over time. However, it has been used for analysing changes in trends as diverse as sales of pipe tobacco [30]; suicide rates [31]; fatal car crashes [32]; workforce growth [33]; and the prevalence of coronary heart disease [34].

Joinpoint regressions can be computationally expensive, and data were therefore transformed into weekly means in order to make analysis tractable. The National Cancer Institute’s Joinpoint Regression Program Version 4.7.0.0 was used for these analyses. Due to the serial nature of the data, adjustments for autocorrelation were made according to [29]. Model selection was conducted by measuring the fit of each model via the calculation of BIC3, a variant of the Bayesian Information Criterion [35, p. 3]. In order to prevent the development of an overfitted model, we elected for a maximum of three joinpoints to be fit to the data, and for a minimum of 8 weeks to occur between joinpoints.

Results

Changes in exposure to loot boxes

Exploratory joinpoint regression was first carried out on the relationship between time and the percentage of individuals in the sample who played games which featured loot boxes. Exposure to loot boxes was initially estimated at 4.2% of the sample in 22^nd-26^th March 2010, rising to 71.2% of the sample by 16th-22nd April 2019. Results indicated that the best-fitting model (BIC3 = 2.63) contained two joinpoints: 1^st-8^th January 2012, and 12^th-19^th March 2014.

Exposure first increased at an average annual rate of 5.3%, from 4.2% at the beginning of observation to 14.0% in the period of 1^st-8^th January 2012 (β = 0.10, t = 2.82, p = 0.004). At this point, the trend increased significantly in steepness (change in β = 0.28, t = 6.50, p<0.001) to an average annual increase of 20.3% (β = 0.39, t = 15.54, p<0.001). Finally, at the second inflection point during 12^th-19^th March 2014, exposure was estimated at 59.4%. At this point, the trend in the data became significantly more shallow (change in β = -0.34, t = -12.96, p<0.001). This led to a more gradual rise in exposure to 71.2% by 16th-22nd April 2019 at an average annual increase of 2.0% (β = 0.04, t = 4.78, p<0.001).

Changes in exposure to pay to win microtransactions

Joinpoint regression was then carried out on the relationship between time and the percentage of individuals in the sample who played games with pay to win features. Exposure to pay to win features was initially estimated at 5.0% of the sample, rising to 15.9% of the sample by 16th-22nd April 2019. Results indicated that the best-fitting model (BIC3 = 1.51) contained a single joinpoint during 12^th-19^th November 2015.

Exposure first increased at an average annual rate of 2.1%, from 5.0% during 22^nd-26^th March 2010 to 17.3% during 12^th-19^th November 2015 (β = 0.04, t = 10.12, p<0.001). At the inflection point of 12^th-19^th November 2015, this trend decreased significantly in steepness (change in β = -0.04, t = -5.40, p<0.001) to an average annual rate that was not significantly different from zero (β = -0.008, t = -0.99, p = 0.32).

Changes in exposure to cosmetic microtransactions

Joinpoint regression was finally carried out on the relationship between time and the percentage of individuals in the sample who played games which featured cosmetic microtransactions. Exposure to cosmetic microtransactions was initially estimated at 8.3% of the sample during 22^nd-26^th March 2010, rising to 85.8% of the sample by 16th-22nd April 2019. Results indicated that the best-fitting model (BIC3 = 2.34) contained two joinpoints: 12^th-19^th February 2012, and 20^th-27^th August 2013.

Exposure first increased at an average annual rate of 7.7%, from 8.3% at 22^nd-26^th March 2010 to 23.4% at 12^th-19^th February 2012 (β = 0.149, t = 5.13, p<0.001). At the first inflection point during 12^th-19^th February 2012, this trend increased significantly in steepness (change in β = 0.40, t = 9.02, p<0.001) to an average annual increase of 28.9% (β = 0.555, t = 16.18, p<0.001), leading to an estimated exposure of 67.8% during 20^th-27^th August 2013. At this point, the trend in the data became significantly more shallow (change in β = -0.49, t = -14.15, p<0.001). This led to a more gradual rise in exposure to 85.8% at 16^th–22^nd April 2019 at an average annual increase of 3.1% (β = 0.06, t = 8.84, p<0.001).

The resulting models from all joinpoint regression analyses are shown below as Fig 1.

Discussion

These results corroborate reports of an overall growth in loot boxes and cosmetic microtransactions in the period 2010–2019. At the beginning of the period, only a small minority of gamers were exposed to these features: 5.3% and 8.3% of the sample respectively. However, by the end of the studied period, the majority of gamers were playing games that featured both loot boxes (71.2%) and cosmetic microtransactions (85.8%). This does not contradict statements by games industry representatives that loot boxes only appear in a minority of games: After all, a mere 75 of the 463 games analysed during this study contained loot boxes. However, they do suggest that the games which do contain loot boxes such as DOTA 2 and Player Unknown’s Battlegrounds may be so popular that, whilst the minority of games may have loot boxes, the majority of gamers are exposed to this feature.

It is important to note that the data under test also provides no evidence of a diminishment in the exposure to either loot boxes or cosmetic microtransactions: None of the regression analyses within any joinpoint model contained a negative coefficient. However, they do suggest that the growth of both cosmetic microtransactions and loot boxes have decelerated in recent years.

The majority of growth in exposure to loot boxes was modelled as taking place between January 2012 and March 2014. During this period exposure to loot boxes increased at a rate of 20.38% per year to a point where more than half of the sample played games with loot boxes. Similarly, between February 2012 and August 2013, exposure to cosmetic microtransactions increased at a rate of 28.9% per year to the point where more than two-thirds of the sample played games with cosmetic microtransactions. However, immediately after these rapid periods of growth, the increase in exposure to both these features dropped significantly to relatively low rates: 2.0% and 3.1% per year respectively. These low rates remained in place for the subsequent 5–6 years.

Some might be surprised by the growth in loot box exposure occurring as early as 2012–2014. They may assume that increases in loot box exposure occurred much later in time. For example, in [16], researchers state that “there were as many games released with loot boxes in 2016–2017 as there were before this time”, suggesting that loot box exposure may similarly experience rapid increases after 2016. The novelty of the result observed here may be due to two factors. The first is that we are measuring exposure rather than prevalence: It may well be the case that changes in the number of games that contain loot boxes occur differently to changes in the number of players who are exposed to loot boxes.

However, another explanation for this difference in analysis is possible: data quality. The statement made above was based on scrutinising a list of games containing loot boxes that was collated and sourced from the video game journalism site Giant Bomb. Said data is subject to various forms of bias and inaccuracy. Furthermore, the underlying processes that are used to generate said data are not available for public scrutiny, and may not conform to scientific ideals. The data presented here is superior in this regard. Further work to determine the changing prevalence rates (as well as exposure rates) of loot boxes over time are necessary.

Exposure to pay to win microtransactions appeared to change in a somewhat different manner to the features outlined above. Whilst loot boxes and cosmetic microtransaction growth was characterised by a sharp increase leading to a slow period of gradual growth, pay to win microtransactions did not experience a similar temporary acceleration. Instead, exposure was modelled as rising at a gradual rate of only 2.1% per year from the beginning of observation until 12^th-19^th November 2015, at which point this rate declined (p<0.001) to an increase that was not significantly different from zero (p = 0.32). Consequently, by the end of the sampled period, only 15.9% of the sample were playing games that featured pay to win microtransactions.

Limitations

The analyses presented here are limited in several ways. The dataset used captures the data of a large number of individuals: Indeed, an average of over 4 million players were recorded each day within our dataset by the conclusion of the studied time period. However, it is important to note that this data represents the players of only the 463 most popular games on Steam. The data of all less-popular games are is therefore not included in this dataset, and it is likely that these games may have a different distribution of features to the most popular games on the market.

Additionally, each game was coded as containing a specific feature if it contained that feature at the time of coding. Theoretically, a game may have only introduced a feature such as cosmetic microtransactions in 2017 or 2018. Yet, when coding took place, all datapoints for that game would be coded as coming from a game which contained such a feature. If this is the case, the models produced below could underestimate the size of increases in exposure within the sample. Furthermore, it is also possible that games in the sample had previously contained loot boxes, and then subsequently removed them. These games would be coded as not containing loot boxes, and their presence in the sample might lead to the overestimation of increases in exposure to loot boxes.

Finally, and most importantly, this dataset consists only of information about desktop games available via the Steam marketplace. It is therefore unable to provide information about the exposure to cosmetic microtransactions, loot boxes, and pay to win features on other platforms such as mobile devices.

One must also note that this data cannot make any claims about the number of players who actually purchased microtransactions of any kind; rather, it speaks to the frequency with which these features appeared in popular games, and the proportion of gamers who are exposed to these features in the games they play.

A final limitation of this study concerns the joinpoint analysis that was used. Joinpoint analyses are able to establish both when changes in the slope of a regression line occurred, and the steepness of a slope after each change-point. This makes them appropriate to the aims of this project: They allow researchers to understand how exposure to various in-game features has changed in specific ways over the past decade. Here, for example, they are used to estimate a rate of increase in exposure between specific dates. However, their utility in understanding other features of exposure is limited: They are not commonly used as predictive models to understand future levels of exposure; and their ability to estimate the shape of trends is limited: They cannot be used to understand if, for example, increases in exposure fit an exponential curve. These are interesting and important analyses for future researchers to conduct. All data associated with our analyses are publicly available, and it is our hope that future work will address these questions.

Similarly, the process of joinpoint regression necessitates researchers defining specifc apriori constraints on their statistical models. For example, the regression undertaken here was constrained to contain a maximum of three joinpoints. It should be noted that these constraints were informed by standard practice in the literature: For example, it is a common strategy to allow a maximum of three joinpoints to occur within a model, presumably in order to resolve tensions between computational efficiency, overfitting, and model accuracy [36–38]. A sceptic may note that this analytic strategy may allow for situations of analytic flexibility: One might receive different results by specifying four, or five, or two joinpoints. Further work may focus on determining the impact of model constraint decisions on analytic outcomes: Our data is open and available with no reservations should other parties wish to reanalyse it for this purpose, via, for example, a form of multiverse analysis.

Conclusions

The exploratory analysis presented above suggests that pay to win microtransactions continue to be an uncommon feature of desktop video games. Increases in exposure to this feature appeared to only gradually rise from 2010 onwards, and to plateau in 2015, leading to relatively low levels of exposure in 2019.

By contrast, cosmetic microtransactions and loot boxes appear to be present in games played by the majority of desktop gamers within the sample. Over 70% of gamers played a game with loot boxes in by the end of the studied period; over 80% played a game with cosmetic microtransactions. This increase in exposure does not appear recent: Indeed, the data suggests that these features may have risen to a dominant position in desktop games as early as 2014.

Academics and policymakers have expressed interest and concern in the potential consequences of the incorporation of the features outlined above in modern video games. Recent reports have suggested that loot boxes may recently have experienced either a decline in popularity, or a rise in popularity. This study instead suggests that, at least on desktop platforms, gamers experienced a sudden increase in exposure to both loot boxes and cosmetic microtransactions during approximately 2012–2014, followed by a period of steady and gradual growth.

Data Availability

All relevant data are held openly on the OSF repository located at https://osf.io/wpqx7/.

Funding Statement

Nick Ballou is a PhD student whose doctoral training is funded by the EPSRC Centre for Doctoral Training in Intelligent Games & Game Intelligence (IGGI).

References

1.Lizardi R., ‘DLC: Perpetual commodification of the video game’, Democr. Commun., vol. 25, no. 1, 2012. [Google Scholar]
2.Consalvo M. and Paul C., ‘Paying to play: The evolving structure of game pricing and industry legitimacy’, AoIR Sel. Pap. Internet Res., vol. 5, 2015. [Google Scholar]
3.Tomić N. Z., ‘economic Model of Microtransactions in video Games’, J. Econ. Sci. Res. Vol., vol. 1, no. 01, 2018. [Google Scholar]
4.Marder B., Gattig D., Collins E., Pitt L., Kietzmann J., and Erz A., ‘The Avatar’s new clothes: Understanding why players purchase non-functional items in free-to-play games’, Comput. Hum. Behav., vol. 91, pp. 72–83, 2019. [Google Scholar]
5.Reza A., Chu S., Khan Z., Nedd A., Castillo A., and Gardner D., ‘Skins for Sale: Linking Player Identity, Representation, and Purchasing Practices’, in Information in Contemporary Society, Cham, 2019, pp. 124–131, 10.1007/978-3-030-15742-5_11 [DOI] [Google Scholar]
6.Davidovici-Nora M., ‘Innovation in business models in the video game industry: Free-To-Play or the gaming experience as a service’, Comput. Games J., vol. 2, no. 3, pp. 22–51, 2013. [Google Scholar]
7.Heimo O. I., Harviainen J. T., Kimppa K. K., and Mäkilä T., ‘Virtual to Virtuous Money: A Virtue Ethics Perspective on Video Game Business Logic’, J. Bus. Ethics, vol. 153, no. 1, pp. 95–103, Nov. 2018, 10.1007/s10551-016-3408-z [DOI] [Google Scholar]
8.Alha K., Koskinen E., Paavilainen J., Hamari J., and Kinnunen J., ‘Free-to-play games: Professionals’ perspectives’, Proc. Nord. DiGRA, vol. 2014, 2014. [Google Scholar]
9.Larche C. J., Musielak N., and Dixon M. J., ‘The Candy Crush Sweet Tooth: How “Near-misses” in Candy Crush Increase Frustration, and the Urge to Continue Gameplay’, J. Gambl. Stud., vol. 33, no. 2, pp. 599–615, Jun. 2017, 10.1007/s10899-016-9633-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.‘Yves Guillemot: “We have no pay-to-win elements in our games”‘, GamesIndustry.biz [Online]. Available: https://www.gamesindustry.biz/articles/2019-10-24-yves-guillemot-we-have-no-pay-to-win-elements-in-our-games. [Accessed: 27-Oct-2019].
11.‘A Guide To The Endless, Confusing Star Wars Battlefront II Controversy’, Kotaku [Online]. Available: https://kotaku.com/a-guide-to-the-endless-confusing-star-wars-battlefront-1820623069. [Accessed: 01-Nov-2019].
12.Zendle D., Meyer R., Waters S., and Cairns P., ‘The prevalence of loot boxes in mobile and desktop games’, Addiction, January 2020. [DOI] [PubMed] [Google Scholar]
13.‘Immersive and addictive technologies report published—News from Parliament’, UK Parliament [Online]. Available: https://www.parliament.uk/business/committees/committees-a-z/commons-select/digital-culture-media-and-sport-committee/news/immersive-technology-report-17-19/. [Accessed: 26-Feb-2020].
14.‘Oral evidence—Immersive and addictive technologies—22 Jul 2019’. [Online]. Available: http://data.parliament.uk/writtenevidence/committeeevidence.svc/evidencedocument/digital-culture-media-and-sport-committee/immersive-and-addictive-technologies/oral/103954.html. [Accessed: 11-Sep-2019].
15.Juniper Research, ‘In-Game Gambling ~ The Next Cash Cow for Publishers; ’, April-2018. [Online]. Available: https://www.juniperresearch.com/document-library/white-papers/in-game-gambling-~-the-next-cash-cow. [Accessed: 15-Jul-2018]. [Google Scholar]
16.Drummond A. and Sauer J. D., ‘Video game loot boxes are psychologically akin to gambling’, Nat. Hum. Behav., pp. 530–532, Jun. 2018, 10.1038/s41562-018-0360-1 [DOI] [PubMed] [Google Scholar]
17.Zendle D. and Bowden-Jones H., ‘Loot boxes and the convergence of video games and gambling’, Lancet Psychiatry, vol. 6, no. 9, pp. 724–725, 2019. [DOI] [PubMed] [Google Scholar]
18.Brooks G. A. and Clark L., ‘Associations between loot box use, problematic gaming and gambling, and gambling-related cognitions’, Addict. Behav., 2019. [DOI] [PubMed] [Google Scholar]
19.Zendle D., Meyer R., and Over H., ‘Adolescents and loot boxes: links with problem gambling and motivations for purchase’, R. Soc. Open Sci., vol. 6, no. 6, p. 190049, Jun. 2019, 10.1098/rsos.190049 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Zendle D. and Cairns P., ‘Video game loot boxes are linked to problem gambling: Results of a large-scale survey’, PLoS One, vol. 13, no. 11, p. e0206767, 2018. 10.1371/journal.pone.0206767 [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Wright M., ‘Video gamers will be spending $50 billion on “gambling-like” loot box features by 2022, according to analysts’, The Telegraph, 17-April-2018. [Google Scholar]
22.Davenport J., ‘Battle passes are replacing loot boxes, but they’re not necessarily a better deal’, PC Gamer, 05-July-2018. [Online]. Available: https://www.pcgamer.com/uk/battle-passes-are-replacing-loot-boxes-but-theyre-not-necessarily-a-better-deal/. [Accessed: 27-Oct-2019]. [Google Scholar]
23.Boland H., ‘British children spending £270m a year on mobile game “loot boxes” and in-app purchases”‘, The Telegraph, 28-August-2019. [Google Scholar]
24.Brunt C. S., King A. S., and King J. T., ‘The influence of user-generated content on video game demand’, J. Cult. Econ., pp. 1–22, 2019. [Google Scholar]
25.Lin D., Bezemer C.-P., and Hassan A. E., ‘An empirical study of early access games on the Steam platform’, Empir. Softw. Eng., vol. 23, no. 2, pp. 771–799, 2018. [Google Scholar]
26.‘Home · SteamDB’, Steam Database [Online]. Available: https://steamdb.info/. [Accessed: 02-Nov-2019].
27.Saldana J., The Coding Manual for Qualitative Researchers. SAGE, 2015. [Google Scholar]
28.Gwet K. L., Handbook of Inter-Rater Reliability, 4th Edition: The Definitive Guide to Measuring The Extent of Agreement Among Raters. Advanced Analytics, LLC, 2014. [Google Scholar]
29.Kim H.-J., Fay M. P., Feuer E. J., and Midthune D. N., ‘Permutation tests for joinpoint regression with applications to cancer rates’, Stat. Med., vol. 19, no. 3, pp. 335–351, 2000. [DOI] [PubMed] [Google Scholar]
30.Morris D. S. and Tynan M. A., ‘Fiscal and Policy Implications of Selling Pipe Tobacco for Roll-Your-Own Cigarettes in the United States’, PLOS ONE, vol. 7, no. 5, p. e36487, May 2012, 10.1371/journal.pone.0036487 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Lee S.-U., Park J.-I., Lee S., Oh I.-H., Choi J.-M., and Oh C.-M., ‘Changing trends in suicide rates in South Korea from 1993 to 2016: a descriptive study’, BMJ Open, vol. 8, no. 9, p. e023144, September 2018, 10.1136/bmjopen-2018-023144 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Nakahara S., Katanoda K., and Ichikawa M., ‘Onset of a declining trend in fatal motor vehicle crashes involving drunk-driving in Japan’, J. Epidemiol., p. JE20120134, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Gruca T. S., Nelson G. C., Thiesen L., Asprey D. P., and Young S. G., ‘The workforce trends of physician assistants in Iowa (1995–2015)’, PLOS ONE, vol. 13, no. 10, p. e0204813, October 2018, 10.1371/journal.pone.0204813 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.O’Flaherty M., Allender S., Taylor R., Stevenson C., Peeters A., and Capewell S., ‘The decline in coronary heart disease mortality is slowing in young adults (Australia 1976–2006): a time trend analysis’, Int. J. Cardiol., vol. 158, no. 2, pp. 193–198, 2012. 10.1016/j.ijcard.2011.01.016 [DOI] [PubMed] [Google Scholar]
35.‘BIC3’, Joinpoint Help System 4.7.0.0 [Online]. Available: https://surveillance.cancer.gov/help/joinpoint/setting-parameters/method-and-parameters-tab/model-selection-method/bic-3. [Accessed: 24-Oct-2019].
36.Vujcic I. S., Sipetic S. B., Dubljanin E. S., and Vlajinac H. D., ‘Trends in mortality rates from coronary heart disease in Belgrade (Serbia) during the period 1990–2010: a joinpoint regression analysis’, BMC Cardiovasc. Disord., vol. 13, no. 1, p. 112, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Thomas K., Chang S.-S., and Gunnell D., ‘Suicide epidemics: the impact of newly emerging methods on overall suicide rates-a time trends study’, BMC Public Health, vol. 11, no. 1, p. 314, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Nistal-Nuño B., ‘Joinpoint regression analysis to evaluate traffic public health policies by national temporal trends from 2000 to 2015’, Int. J. Inj. Contr. Saf. Promot., vol. 25, no. 2, pp. 128–133, 2018. 10.1080/17457300.2017.1341937 [DOI] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0232780.r001

Decision Letter 0

José C Perales

28 Jan 2020

PONE-D-19-30715

The changing face of desktop video game monetisation: An exploration of trends in loot boxes, pay to win, and cosmetic microtransactions in the most-played Steam games of 2010-2019

PLOS ONE

Dear Dr. Zendle,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Editor comments:

I have now the reviews on your manuscript from three experts on the topic. In spite of the differences in the tone of their general assessments, the three reports are mostly coincident in content, and converge in presenting some concerns that should be addressed before the paper is considered suitable for publication.

The first of them has to do with the very definition of pay-to-win, cosmetic microtransactions, and loot boxes. As pointed out by R2 and R3, loot boxes can contain items that can confer or not competitive advantage to the player who purchases them, so some overlap exists between loot boxes and the other two categories. Please provide the necessary information on how this overlap was resolved (as well as on the other definitional aspects mentioned by the three reviewers).

The second important issue regards the use of “prevalence”. In my view (and R3’s), prevalence seems to refer to the number of games that contain a feature, whereas “exposure” better describes how many people are exposed to such a feature. Actually, I completely agree with R3 that the two analyses are interesting enough to be presented together. (In terms of policy, the implications of an increasing number of games including certain feature are very different to the ones of an increasing number of people purchasing and playing games including that feature). And, in any case, a systematic and transparent terminology should be used throughout the manuscript.

In relation to this (sorry if I missed something here), I guess there are games for which features of interests (i.e. microtransactions) were added or removed during the period under scrutiny. Did that actually occur? And, if it did, how was it recorded and taken into account for analyses?

Finally, although not mentioned by the reviewers, I suggest to better justify the statistical approach. Why was joinpoint analysis used instead of trying to fit some continuous function? As far as I know, joinpoint regression is aimed at detecting time points in which trends change, so they can be interpreted in relation to certain events (e.g. policy changes, or new legislations). I understand that trying explain post-hoc what happened at such point would be highly speculative. Yet, in the absence of prior hypotheses regarding trend changes, the use of this analysis (and the a priori decision to limit the number of joinpoints to 3) may seem arbitrary.

We would appreciate receiving your revised manuscript by Mar 13 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

José C. Perales

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. In your ethics statement in the online submission information and in the manuscript, please include the full name of the department that issued the ethics waiver.

3. Please amend your list of authors on the manuscript to ensure that each author is linked to an affiliation. Authors’ affiliations should reflect the institution where the work was done (if authors moved subsequently, you can also list the new affiliation stating “current affiliation:….” as necessary).

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This was a very clearly written article and the data with an interesting application of joinpoint regression that I had not seen before. I have only a couple of observations:

1) In the 4 paragraph (introduction) you note that "Loot Boxes are items or bonuses in video games that players can buy with real-world money...." However, Loot Boxes can also refer to in-game items that are rewarded from gameplay or non-monetary points accumulated in a game. It would be helpful if you simply noted your more restrictive definition of loot boxes.

2) When quoting growth rates throughout the document, you use 2 significant digits (e.g., 5.31%). Given the nature of the analyses, I don't think the second digit is at all reliable, and thus indicates a level of precision in your estimates that likely does not exist. I suggest - and this is only a suggestion - that it would be more appropriate to only use 1 significant digit (e.g., 5.3%).

Reviewer #2: The paper seeks to examine the prevalence rate of two types of purchasable virtual items, so-called cosmetic and pay-to-win items, and a monetization technique colloquially known as loot boxes.

The paper examines an interesting and important topic in game studies. However, I have some major concerns with the paper, one is methodological and one in ontological. First of all, the authors several times distinguish between cosmetic microtransactions, pay-to-win microtransactions, and loot boxes. This gives the impression that the authors understand these to be distinct objects. However, since loot boxes can contain both cosmetic and pay-to-win objects this is misleading to the reader. Pay-to-win items and cosmetic items, it seems to me that loot boxes are not as much items as a procedure for doling out rewards or products in a transaction. I'd like the authors to acknowledge and reflect on this. Or alternatively tell me why I am wrong in my view.

Second of all, it is not made clear how the researchers have obtained a deep understanding of the vast number of games that they have scored. I am not convinced by the paper that the authors are familiar enough with the games that they score for the scores to be meaning full. Even if every game can be played through relatively quickly it would still be a massively time-consuming endeavor to play through all of them. Is the rating process based on firsthand knowledge of the games or is there a database somewhere that they are using?

Thirdly, I think the authors should be more precise in their definition of what loot boxes are. If we take packs of footballers purchased by the player in Fifa as an example, I’d like to know if the authors categorize these as loot boxes or not. The packs are the same in all countries, but in some they can be purchase with real money and in some they cannot. Following the authors definition these same packs could then be categorized as both loot boxes and not loot boxes. This, I assume is true of all games. This, I think, requires the authors to be much more transparent about their methodology and results.

Two minor problems here at the end:

1) In the introduction it seems that the authors forgot the part of history where games as a service emerged (i.e. subscription-based monetization strategies).

2) The misunderstanding about DLC and whether to code that as pay-to-win or cosmetic microtransactions leave me questioning the competence of the raters. The paper reads as follows: “After this round of coding, it emerged that disagreements in coding may have been due to a lack of clarity about whether downloadable content (DLC) such as expansion packs should be classified as either pay to win or cosmetic microtransactions. In order to resolve this, it was agreed that cosmetic and pay to win microtransactions would be classified as in-game items and rewards that are purchasable with real-world money but do not add substantial additional game content.”

That such an argument could even take place makes me question the raters game literacy or knowledge of games, which is problematic because the raters rating of a massive amount of games is the foundation of the paper.

All in all, I think it is an interesting paper, and I wish I could have given it a better review, but I think the paper is not fit for publication in its current state. I tried to access the data that is made available online, but found the data in the files to be really hard to parse. Maybe the files are just corrupt or otherwise damaged.

Reviewer #3: This is a well-written manuscript reporting an interesting piece of research. I enjoyed reading it, and believe these data will make a valuable contribution to the literature. The introduction provides a nice, concise introduction to the issue based on the extant literature, and a solid rationale for this work. Essentially, much is being of made of the potential risks associated with predatory monetization practices, but the actual prevalence of these practices is disputed. A systematic empirical examination of the prevalence of monetization mechanisms in video games will help researchers, policy-makers, and gamers to better appreciate the potential for risk posed by these mechanisms. As I say, I think this research is valuable. However, below I outline some areas where, based on my reading, the manuscript might benefit from further clarification. Importantly, I don’t see any of these issues as fatal.

Introduction

The taxonomies presented to characterize micro-transactions could be clearer. There seem to be multiple dimensions at play here. First, are the virtual items obtained purely aesthetic, or do they alter gameplay and confer an in-game advantage? Second, are they purchased directly, or obtained randomized reward mechanisms (e.g., loot boxes)? Although items that provide competitive boosts and are purchased directly seem to clearly fall under the “pay-to-win” heading, I’m not sure items obtained from loot boxes would, even if they confer a competitive advantage. Admittedly, this might be a purely subjective distinction. However, I wonder if both the function of the item and the method of acquisition are likely to be important to understanding whether the microtransaction in question constitutes a pay-to-win mechanism. Similarly, when considering “gambling-like entrapment”, it seems important to separate the nature of the item from the nature of the reward delivery mechanism (i.e., randomized outcome vs. direct purchase). This becomes clearer in the “present research” section – where microtransactions are characterized as loot boxes, pay to win, or cosmetic – but I think this issue could be a little more clearly distilled in the introduction.

Method

* The method for acquiring data was generally clearly articulated, and the criteria for selecting the games included seems sensible. How did reviewers code for the presence or absence of microtransactions? Did they source data from Steam, or from game review platforms, etc.?

* Interested readers might benefit from further technical details (e.g., in supplemental files) relating to how data were obtained from the SteamDB website.

Results

* I’m not familiar with joinpoint regression, but the analysis was described clearly from a conceptual standpoint and seems appropriate for the task at hand.

* Having said that, I was a little surprised by the analysis reported. To estimate the prevalence of microtransactions in games and changes in this prevalence over time, I expected to see a comparison of the percentage of popular games featuring and not featuring microtransactions on a year-by-year basis. Admittedly, this is a comparatively unsophisticated analysis, but it seems to get at the prevalence issue. The analyses reported seem to speak to a separate issue: The percentage of players who are playing games with microtransactions. This is potentially an interesting question in itself, and may relate to prevalence of engagement with these mechanisms (though it is a proxy at best: as the authors note in the discussion, playing a game with microtransactions and engaging with microtransactions are not the same thing), but it doesn’t seem to speak to the prevalence of microtransactions themselves. I might be wrong here but, if so, I think the authors could be clearer about how the chosen analysis speaks to the prevalence of these mechanisms in the manner discussed in the introduction.

* Nonetheless, the analyses presented are clearly reported, and augmented nicely by the data in Figure 1.

Discussion

* Again, I’m not sure these analyses show “an overall growth in loot boxes and cosmetic microtransactions”. Instead, they seem to show an increase in the percentage of players in the sample who played games featuring microtransactions. Again, this seems to speak to player engagement with games housing these features, rather than the prevalence of the features themselves. This is hinted at by the authors acknowledgment that, even though most of the gamers in the sample were playing games that included loot boxes, only 75 of 463 games reviewed contained loot boxes. Again, apologies if I’ve misunderstood what these data are telling me.

* Nonetheless, the more basic interpretations of the data are sound: engagement with games that feature microtransactions and loot boxes increased notably over the period of time for which data were analyzed.

* It interesting to note that, perhaps counter to many people’s perceptions, the most rapid period of growth (according to the data presented here) was between 2012 and 2014 (cf. the more recent examples that have captured public and media attention). Though, once again, this potentially highlights a divergence from the issue of prevalence: Drummond & Sauer (2018) note that were as many games released with loot boxes in 2016–2017 as there were before this time suggesting that, for console games at least, the prevalence of loot box mechanisms increased most rapidly in that time period (cf. the 2012-2014 evidenced in the current data). I wonder if this difference reflects (a) differences in console vs. desktop gaming or (b) differences in the way prevalence is estimated?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2020 May 7;15(5):e0232780. doi: 10.1371/journal.pone.0232780.r002

Author response to Decision Letter 0

2 Mar 2020

Dear Dr. Perales,

We thank both yourself and the reviewers for their detailed remarks. We provide a point-by-point response to each comment below.

“I have now the reviews on your manuscript from three experts on the topic. In spite of the differences in the tone of their general assessments, the three reports are mostly coincident in content, and converge in presenting some concerns that should be addressed before the paper is considered suitable for publication.

We have extended our method section to describe how overlap between loot boxes, pay to win, and cosmetic microtransactions was considered within our analysis. Other definitional questions touched on by the editor have been addressed through additions to our method and introduction. These are detailed below in the rest of this rebuttal, alongside relevant reviewer comments.

“The second important issue regards the use of “prevalence”. In my view (and R3’s), prevalence seems to refer to the number of games that contain a feature, whereas “exposure” better describes how many people are exposed to such a feature. Actually, I completely agree with R3 that the two analyses are interesting enough to be presented together. (In terms of policy, the implications of an increasing number of games including certain feature are very different to the ones of an increasing number of people purchasing and playing games including that feature). And, in any case, a systematic and transparent terminology should be used throughout the manuscript.”

We have adjusted our terminology throughout the manuscript in line with this comment. We now consistently refer to exposure as the number of gamers playing games with a feature; and prevalence as the number of games with that feature.

“In relation to this (sorry if I missed something here), I guess there are games for which features of interests (i.e. microtransactions) were added or removed during the period under scrutiny. Did that actually occur? And, if it did, how was it recorded and taken into account for analyses?”

The addition or removal of features during the time-period under test was not taken into account in our analyses. Our method has been augmented to clarify why this was the case; our discussion has described how this may affect our analyses, and suggested further work on the topic.

“Finally, although not mentioned by the reviewers, I suggest to better justify the statistical approach. Why was joinpoint analysis used instead of trying to fit some continuous function? As far as I know, joinpoint regression is aimed at detecting time points in which trends change, so they can be interpreted in relation to certain events (e.g. policy changes, or new legislations). I understand that trying explain post-hoc what happened at such point would be highly speculative. Yet, in the absence of prior hypotheses regarding trend changes, the use of this analysis (and the a priori decision to limit the number of joinpoints to 3) may seem arbitrary.”

The ‘limitations’ subsection of our manuscript has been extended to better motivate the use of joinpoint analysis and to highlight the open nature of the dataset for future re-analysis and investigation.

“Reviewer #1: This was a very clearly written article and the data with an interesting application of joinpoint regression that I had not seen before. I have only a couple of observations:

Our manuscript has been altered to highlight the definition of loot boxes used here: It should be noted that this is the definition used by the UK government at a recent Select Committee inquiry, so it may come into more common usage.

“2) When quoting growth rates throughout the document, you use 2 significant digits (e.g., 5.31%). Given the nature of the analyses, I don't think the second digit is at all reliable, and thus indicates a level of precision in your estimates that likely does not exist. I suggest - and this is only a suggestion - that it would be more appropriate to only use 1 significant digit (e.g., 5.3%).”

We have adjusted our manuscript as suggested by the reviewer, and 1 significant digit is now used throughout

“Reviewer #2: The paper seeks to examine the prevalence rate of two types of purchasable virtual items, so-called cosmetic and pay-to-win items, and a monetization technique colloquially known as loot boxes.

The manuscript has been augmented to incorporate a more clear delineation between loot boxes, pay to win microtransactions, and cosmetic microtransactions. The definitional points made by R2 have been taken into account.

“Second of all, it is not made clear how the researchers have obtained a deep understanding of the vast number of games that they have scored. I am not convinced by the paper that the authors are familiar enough with the games that they score for the scores to be meaning full. Even if every game can be played through relatively quickly it would still be a massively time-consuming endeavor to play through all of them. Is the rating process based on firsthand knowledge of the games or is there a database somewhere that they are using?”

The manuscript has been updated to more fully describe the process that the authors went through in order to establish the presence of loot boxes in each game.

“Thirdly, I think the authors should be more precise in their definition of what loot boxes are. If we take packs of footballers purchased by the player in Fifa as an example, I’d like to know if the authors categorize these as loot boxes or not. The packs are the same in all countries, but in some they can be purchase with real money and in some they cannot. Following the authors definition these same packs could then be categorized as both loot boxes and not loot boxes. This, I assume is true of all games. This, I think, requires the authors to be much more transparent about their methodology and results.”

The definition of loot boxes used in this study has been further highlighted, and a brief discussion of the limitations and advantages of the approach to definition used has been incorporated into the manuscript.

“Two minor problems here at the end:

1) In the introduction it seems that the authors forgot the part of history where games as a service emerged (i.e. subscription-based monetization strategies).”

We have augmented our introduction to briefly mention subscription-based monetisation

“2) The misunderstanding about DLC and whether to code that as pay-to-win or cosmetic microtransactions leave me questioning the competence of the raters. The paper reads as follows: “After this round of coding, it emerged that disagreements in coding may have been due to a lack of clarity about whether downloadable content (DLC) such as expansion packs should be classified as either pay to win or cosmetic microtransactions. In order to resolve this, it was agreed that cosmetic and pay to win microtransactions would be classified as in-game items and rewards that are purchasable with real-world money but do not add substantial additional game content.”

Disagreements between coders are extremely common during reliability analyses. We believe there is a danger in reflexively stigmatising such disagreements. We strongly believe that it is important to be transparent about these disagreements, in order to create a manuscript that most accurately reflects the underlying data and analysis. We have augmented our manuscript to describe how common this is, and hope that this discussion will convince the reviewer of the credibility of the research team.

“All in all, I think it is an interesting paper, and I wish I could have given it a better review, but I think the paper is not fit for publication in its current state. I tried to access the data that is made available online, but found the data in the files to be really hard to parse. Maybe the files are just corrupt or otherwise damaged.”

The data associated with this manuscript are freely available on the OSF repository associated with this manuscript, as are the exact scripts used to run the major analyses. We appreciate that these may be difficult to parse, and hence have included a brief document as a key in response to R2’s critique.

“The taxonomies presented to characterize micro-transactions could be clearer. There seem to be multiple dimensions at play here. First, are the virtual items obtained purely aesthetic, or do they alter gameplay and confer an in-game advantage? Second, are they purchased directly, or obtained randomized reward mechanisms (e.g., loot boxes)? Although items that provide competitive boosts and are purchased directly seem to clearly fall under the “pay-to-win” heading, I’m not sure items obtained from loot boxes would, even if they confer a competitive advantage. Admittedly, this might be a purely subjective distinction. However, I wonder if both the function of the item and the method of acquisition are likely to be important to understanding whether the microtransaction in question constitutes a pay-to-win mechanism.”

A common theme of comments from all reviewers is the need for a clearer discussion of what is specifically meant in this manuscript when a game is categorised as containing either pay to win, loot boxes, or cosmetic microtransactions. This point is well-taken, and a detailed discussion that addresses R3’s questions has been incorporated into the manuscript.

Similarly, when considering “gambling-like entrapment”, it seems important to separate the nature of the item from the nature of the reward delivery mechanism (i.e., randomized outcome vs. direct purchase). This becomes clearer in the “present research” section – where microtransactions are characterized as loot boxes, pay to win, or cosmetic – but I think this issue could be a little more clearly distilled in the introduction.

The manuscript has been augmented with additional information clarifying the potential role of gambling-like entrapment in in-game spending.

“Method

A more clear and detailed description of how each game was coded has been incorporated into our method section in response to reviewer feedback.

* Interested readers might benefit from further technical details (e.g., in supplemental files) relating to how data were obtained from the SteamDB website.

Further technical details have been incorporated into the manuscript about how data was collected from the SteamDB website. However, we feel that the reviewer may find our methods for data extraction somewhat plodding and less exciting than they may have imagined!

Results

* I’m not familiar with joinpoint regression, but the analysis was described clearly from a conceptual standpoint and seems appropriate for the task at hand.

“* Having said that, I was a little surprised by the analysis reported. To estimate the prevalence of microtransactions in games and changes in this prevalence over time, I expected to see a comparison of the percentage of popular games featuring and not featuring microtransactions on a year-by-year basis. Admittedly, this is a comparatively unsophisticated analysis, but it seems to get at the prevalence issue. The analyses reported seem to speak to a separate issue: The percentage of players who are playing games with microtransactions. This is potentially an interesting question in itself, and may relate to prevalence of engagement with these mechanisms (though it is a proxy at best: as the authors note in the discussion, playing a game with microtransactions and engaging with microtransactions are not the same thing), but it doesn’t seem to speak to the prevalence of microtransactions themselves. I might be wrong here but, if so, I think the authors could be clearer about how the chosen analysis speaks to the prevalence of these mechanisms in the manner discussed in the introduction.”

* Nonetheless, the analyses presented are clearly reported, and augmented nicely by the data in Figure 1.

Discussion

The reviewer is correct in their inferences, and their points are well-taken. In response to this comment (and aligned editorial feedback), we have revised the manuscript to frame our analyses as a description of ‘exposure’ rather than ‘prevalence’.

“* Nonetheless, the more basic interpretations of the data are sound: engagement with games that feature microtransactions and loot boxes increased notably over the period of time for which data were analyzed.

The reviewer raises interesting points. A treatment of the points raised – including specific reference to Drummond and Sauer’s points - has been integrated into our discussion.

PLoS One. doi: 10.1371/journal.pone.0232780.r003

Decision Letter 1

José C Perales

11 Apr 2020

PONE-D-19-30715R1

The changing face of desktop video game monetisation: An exploration of exposure to loot boxes, pay to win, and cosmetic microtransactions in the most-played Steam games of 2010-2019

PLOS ONE

Dear Dr. Zendle,

Thank you for submitting your revised manuscript to PLOS ONE.

As you can see in the attached reports, the reviewers are mostly satisfied with the level of detail with which you have considered their comments and suggestions.

Only one of them still finds your definition of loot boxes not precise enough fot an academic context. Please try to address that remaining concern etither making minor changes in your manuscript or in your response letter. In principle, if I find your reply convincing enough, no further review rounds will be required.

We would appreciate receiving your revised manuscript by May 26 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

We look forward to receiving your revised manuscript.

Kind regards,

José C. Perales

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

Reviewer #3: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: I Don't Know

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Reviewer #1: My comments and suggestions have been adequately addressed. The problem with definitions of loot boxes has been addressed, and the suggestion to reduce significant digits to 1 was also accepted.

Reviewer #2: I thank the authors for their replies to our comments. I am satisfied that they have made the needed corrections in most cases. Furthermore, I am impressed that the researchers have downloaded and played such an enormous amount of games. Cudos.

The one exception is their definition of loot boxes, which is still think is too vague and ambiguous for an academic paper. The authors still do not account for the fact that the same loot box can be purchased with fiat currency or in-game currency accrued through play. Instead, they seem to argue that the same item (e.g. a card pack) in digital games are loot boxes if purchased with fiat currency and not loot boxes if purchased through play.

I am not convinced that the definition provided by the UK Parliament is of much use in an academic context. This would not be problematic if the authors did not liken loot boxes to slot machines. If the authors believe that digital collectible card games are gambling do the authors then also believe that non-digital collectible card games are gambling? I’m assuming this would go against the view of the UK Parliament.

I’m sure this line of argument seems pedantic to the authors, but I think it is disingenuous to imply that loot boxes are simply slot machines in games. Surely there are more nuances than that? Otherwise, how could so many other countries in the world come to the conclusion that loot boxes are in fact not gambling? Acknowledging the complexities of the issue would go a long way.

Reviewer #3: PONE-D-19-30715_R1

I was Reviewer 3 on the original manuscript.

The authors have been thorough in their response to the reviews. In particular, the changes to the introduction make for a clearer explanation of cosmetic and pay-to-win microtransactions, and their relationship to loot boxes. In general, as a reader, I appreciated the additional detail relating to how the data were obtained and how they were coded (even if this detail was not necessarily exciting). I liked the manuscript before, and I think it is even better now. It will make a valuable contribution to the literature.

I have nothing further to request from the authors.

As I side note, it was really useful to have the track-changes version of the document available to appreciate the scope of the revisions made.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

PLoS One. 2020 May 7;15(5):e0232780. doi: 10.1371/journal.pone.0232780.r004

Author response to Decision Letter 1

11 Apr 2020

Dear Dr. Perales,

We thank both yourself and the reviewers for their remarks. We provide a response to reviewer 2’s comment below.

Reviewer 2's comments

---------------------------------------------------------

I thank the authors for their replies to our comments. I am satisfied that they have made the needed corrections in most cases. Furthermore, I am impressed that the researchers have downloaded and played such an enormous amount of games. Cudos.

The reviewer makes several points here which we would like to respond to, as they contain some misinterpretation of the work presented in this paper.

--------------------------------------------------------

Firstly, reviewer 2 states that the definition of loot boxes that we use in our manuscript is ‘provided by’ the UK Parliament, and therefore is of little use in an academic context. However, they are incorrect. If they inspect our manuscript, they will see that this definition is taken from a recent article published in Addiction (a scholarly journal). As noted in the manuscript, this academic definition was also *used* by a UK Parliamentary Select Committee to define loot boxes for the purposes of making recommendations regarding regulation. We believe this uptake in both academic and governmental contexts suggests it may have some utility.

Secondly, the reviewer provides the following objection regarding our definition of loot boxes: “I think it is disingenuous to imply that loot boxes are simply slot machines in games”. We have not stated that loot boxes are ‘simply slot machines in games’ in our manuscript, and do not believe it to be the case. This paper does not engage in definitional work of that nature in any way. We are unclear where the reviewer believes we make this argument. The closest that we come to is the text quoted below:

As noted in [16], loot boxes share distinct similarities with gambling. Both when paying for a loot box and when putting money into a slot machine, individuals are wagering something of value on the chance hope of receiving something of greater value. This has led to concerns that engaging with loot boxes may lead to increases in gambling amongst gamers [17]. Evidence for this causal mechanism is unclear. Spending on loot boxes has been repeatedly linked to problem gambling. However, it is uncertain whether this is because loot boxes cause problem gambling, or whether it is because individuals with pre-existing gambling problems spend more money on loot boxes [18]–[20].

We find the text above uncontroversial and accurate. However, in order to assure that misinterpretations such as the one reviewer 2 makes do not occur, we have removed the following sentence from our manuscript: “Both when paying for a loot box and when putting money into a slot machine, individuals are wagering something of value on the chance hope of receiving something of greater value.”

Thirdly, the reviewer states that they are unconvinced that our definition of loot boxes is of use in an academic context for the following reason: “This would not be problematic if the authors did not liken loot boxes to slot machines. If the authors believe that digital collectible card games are gambling do the authors then also believe that non-digital collectible card games are gambling?”.

We are not entirely clear with what reviewer 2 is getting at here. They state that the authors ‘believe that digital collectible card games are gambling’. Yet nowhere in this manuscript have we classified loot boxes as ‘gambling’ or ‘not gambling’. The only paragraph that could even potentially be read this way is the one we quote above. As we have now removed this sentence, we hope that this potential misinterpretation may also now be resolved.

PLoS One. doi: 10.1371/journal.pone.0232780.r005

Decision Letter 2

José C Perales

22 Apr 2020

The changing face of desktop video game monetisation: An exploration of exposure to loot boxes, pay to win, and cosmetic microtransactions in the most-played Steam games of 2010-2019

PONE-D-19-30715R2

Dear Dr. Zendle,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

José C. Perales

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

PLoS One. doi: 10.1371/journal.pone.0232780.r006

Acceptance letter

José C Perales

28 Apr 2020

PONE-D-19-30715R2

The changing face of desktop video game monetisation: An exploration of exposure to loot boxes, pay to win, and cosmetic microtransactions in the most-played Steam games of 2010-2019

Dear Dr. Zendle:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. José C. Perales

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

All relevant data are held openly on the OSF repository located at https://osf.io/wpqx7/.

[pone.0232780.ref001] 1.Lizardi R., ‘DLC: Perpetual commodification of the video game’, Democr. Commun., vol. 25, no. 1, 2012. [Google Scholar]

[pone.0232780.ref002] 2.Consalvo M. and Paul C., ‘Paying to play: The evolving structure of game pricing and industry legitimacy’, AoIR Sel. Pap. Internet Res., vol. 5, 2015. [Google Scholar]

[pone.0232780.ref003] 3.Tomić N. Z., ‘economic Model of Microtransactions in video Games’, J. Econ. Sci. Res. Vol., vol. 1, no. 01, 2018. [Google Scholar]

[pone.0232780.ref004] 4.Marder B., Gattig D., Collins E., Pitt L., Kietzmann J., and Erz A., ‘The Avatar’s new clothes: Understanding why players purchase non-functional items in free-to-play games’, Comput. Hum. Behav., vol. 91, pp. 72–83, 2019. [Google Scholar]

[pone.0232780.ref005] 5.Reza A., Chu S., Khan Z., Nedd A., Castillo A., and Gardner D., ‘Skins for Sale: Linking Player Identity, Representation, and Purchasing Practices’, in Information in Contemporary Society, Cham, 2019, pp. 124–131, 10.1007/978-3-030-15742-5_11 [DOI] [Google Scholar]

[pone.0232780.ref006] 6.Davidovici-Nora M., ‘Innovation in business models in the video game industry: Free-To-Play or the gaming experience as a service’, Comput. Games J., vol. 2, no. 3, pp. 22–51, 2013. [Google Scholar]

[pone.0232780.ref007] 7.Heimo O. I., Harviainen J. T., Kimppa K. K., and Mäkilä T., ‘Virtual to Virtuous Money: A Virtue Ethics Perspective on Video Game Business Logic’, J. Bus. Ethics, vol. 153, no. 1, pp. 95–103, Nov. 2018, 10.1007/s10551-016-3408-z [DOI] [Google Scholar]

[pone.0232780.ref008] 8.Alha K., Koskinen E., Paavilainen J., Hamari J., and Kinnunen J., ‘Free-to-play games: Professionals’ perspectives’, Proc. Nord. DiGRA, vol. 2014, 2014. [Google Scholar]

[pone.0232780.ref009] 9.Larche C. J., Musielak N., and Dixon M. J., ‘The Candy Crush Sweet Tooth: How “Near-misses” in Candy Crush Increase Frustration, and the Urge to Continue Gameplay’, J. Gambl. Stud., vol. 33, no. 2, pp. 599–615, Jun. 2017, 10.1007/s10899-016-9633-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref010] 10.‘Yves Guillemot: “We have no pay-to-win elements in our games”‘, GamesIndustry.biz [Online]. Available: https://www.gamesindustry.biz/articles/2019-10-24-yves-guillemot-we-have-no-pay-to-win-elements-in-our-games. [Accessed: 27-Oct-2019].

[pone.0232780.ref011] 11.‘A Guide To The Endless, Confusing Star Wars Battlefront II Controversy’, Kotaku [Online]. Available: https://kotaku.com/a-guide-to-the-endless-confusing-star-wars-battlefront-1820623069. [Accessed: 01-Nov-2019].

[pone.0232780.ref012] 12.Zendle D., Meyer R., Waters S., and Cairns P., ‘The prevalence of loot boxes in mobile and desktop games’, Addiction, January 2020. [DOI] [PubMed] [Google Scholar]

[pone.0232780.ref013] 13.‘Immersive and addictive technologies report published—News from Parliament’, UK Parliament [Online]. Available: https://www.parliament.uk/business/committees/committees-a-z/commons-select/digital-culture-media-and-sport-committee/news/immersive-technology-report-17-19/. [Accessed: 26-Feb-2020].

[pone.0232780.ref014] 14.‘Oral evidence—Immersive and addictive technologies—22 Jul 2019’. [Online]. Available: http://data.parliament.uk/writtenevidence/committeeevidence.svc/evidencedocument/digital-culture-media-and-sport-committee/immersive-and-addictive-technologies/oral/103954.html. [Accessed: 11-Sep-2019].

[pone.0232780.ref015] 15.Juniper Research, ‘In-Game Gambling ~ The Next Cash Cow for Publishers; ’, April-2018. [Online]. Available: https://www.juniperresearch.com/document-library/white-papers/in-game-gambling-~-the-next-cash-cow. [Accessed: 15-Jul-2018]. [Google Scholar]

[pone.0232780.ref016] 16.Drummond A. and Sauer J. D., ‘Video game loot boxes are psychologically akin to gambling’, Nat. Hum. Behav., pp. 530–532, Jun. 2018, 10.1038/s41562-018-0360-1 [DOI] [PubMed] [Google Scholar]

[pone.0232780.ref017] 17.Zendle D. and Bowden-Jones H., ‘Loot boxes and the convergence of video games and gambling’, Lancet Psychiatry, vol. 6, no. 9, pp. 724–725, 2019. [DOI] [PubMed] [Google Scholar]

[pone.0232780.ref018] 18.Brooks G. A. and Clark L., ‘Associations between loot box use, problematic gaming and gambling, and gambling-related cognitions’, Addict. Behav., 2019. [DOI] [PubMed] [Google Scholar]

[pone.0232780.ref019] 19.Zendle D., Meyer R., and Over H., ‘Adolescents and loot boxes: links with problem gambling and motivations for purchase’, R. Soc. Open Sci., vol. 6, no. 6, p. 190049, Jun. 2019, 10.1098/rsos.190049 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref020] 20.Zendle D. and Cairns P., ‘Video game loot boxes are linked to problem gambling: Results of a large-scale survey’, PLoS One, vol. 13, no. 11, p. e0206767, 2018. 10.1371/journal.pone.0206767 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref021] 21.Wright M., ‘Video gamers will be spending $50 billion on “gambling-like” loot box features by 2022, according to analysts’, The Telegraph, 17-April-2018. [Google Scholar]

[pone.0232780.ref022] 22.Davenport J., ‘Battle passes are replacing loot boxes, but they’re not necessarily a better deal’, PC Gamer, 05-July-2018. [Online]. Available: https://www.pcgamer.com/uk/battle-passes-are-replacing-loot-boxes-but-theyre-not-necessarily-a-better-deal/. [Accessed: 27-Oct-2019]. [Google Scholar]

[pone.0232780.ref023] 23.Boland H., ‘British children spending £270m a year on mobile game “loot boxes” and in-app purchases”‘, The Telegraph, 28-August-2019. [Google Scholar]

[pone.0232780.ref024] 24.Brunt C. S., King A. S., and King J. T., ‘The influence of user-generated content on video game demand’, J. Cult. Econ., pp. 1–22, 2019. [Google Scholar]

[pone.0232780.ref025] 25.Lin D., Bezemer C.-P., and Hassan A. E., ‘An empirical study of early access games on the Steam platform’, Empir. Softw. Eng., vol. 23, no. 2, pp. 771–799, 2018. [Google Scholar]

[pone.0232780.ref026] 26.‘Home · SteamDB’, Steam Database [Online]. Available: https://steamdb.info/. [Accessed: 02-Nov-2019].

[pone.0232780.ref027] 27.Saldana J., The Coding Manual for Qualitative Researchers. SAGE, 2015. [Google Scholar]

[pone.0232780.ref028] 28.Gwet K. L., Handbook of Inter-Rater Reliability, 4th Edition: The Definitive Guide to Measuring The Extent of Agreement Among Raters. Advanced Analytics, LLC, 2014. [Google Scholar]

[pone.0232780.ref029] 29.Kim H.-J., Fay M. P., Feuer E. J., and Midthune D. N., ‘Permutation tests for joinpoint regression with applications to cancer rates’, Stat. Med., vol. 19, no. 3, pp. 335–351, 2000. [DOI] [PubMed] [Google Scholar]

[pone.0232780.ref030] 30.Morris D. S. and Tynan M. A., ‘Fiscal and Policy Implications of Selling Pipe Tobacco for Roll-Your-Own Cigarettes in the United States’, PLOS ONE, vol. 7, no. 5, p. e36487, May 2012, 10.1371/journal.pone.0036487 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref031] 31.Lee S.-U., Park J.-I., Lee S., Oh I.-H., Choi J.-M., and Oh C.-M., ‘Changing trends in suicide rates in South Korea from 1993 to 2016: a descriptive study’, BMJ Open, vol. 8, no. 9, p. e023144, September 2018, 10.1136/bmjopen-2018-023144 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref032] 32.Nakahara S., Katanoda K., and Ichikawa M., ‘Onset of a declining trend in fatal motor vehicle crashes involving drunk-driving in Japan’, J. Epidemiol., p. JE20120134, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref033] 33.Gruca T. S., Nelson G. C., Thiesen L., Asprey D. P., and Young S. G., ‘The workforce trends of physician assistants in Iowa (1995–2015)’, PLOS ONE, vol. 13, no. 10, p. e0204813, October 2018, 10.1371/journal.pone.0204813 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref034] 34.O’Flaherty M., Allender S., Taylor R., Stevenson C., Peeters A., and Capewell S., ‘The decline in coronary heart disease mortality is slowing in young adults (Australia 1976–2006): a time trend analysis’, Int. J. Cardiol., vol. 158, no. 2, pp. 193–198, 2012. 10.1016/j.ijcard.2011.01.016 [DOI] [PubMed] [Google Scholar]

[pone.0232780.ref035] 35.‘BIC3’, Joinpoint Help System 4.7.0.0 [Online]. Available: https://surveillance.cancer.gov/help/joinpoint/setting-parameters/method-and-parameters-tab/model-selection-method/bic-3. [Accessed: 24-Oct-2019].

[pone.0232780.ref036] 36.Vujcic I. S., Sipetic S. B., Dubljanin E. S., and Vlajinac H. D., ‘Trends in mortality rates from coronary heart disease in Belgrade (Serbia) during the period 1990–2010: a joinpoint regression analysis’, BMC Cardiovasc. Disord., vol. 13, no. 1, p. 112, 2013. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref037] 37.Thomas K., Chang S.-S., and Gunnell D., ‘Suicide epidemics: the impact of newly emerging methods on overall suicide rates-a time trends study’, BMC Public Health, vol. 11, no. 1, p. 314, 2011. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0232780.ref038] 38.Nistal-Nuño B., ‘Joinpoint regression analysis to evaluate traffic public health policies by national temporal trends from 2000 to 2015’, Int. J. Inj. Contr. Saf. Promot., vol. 25, no. 2, pp. 128–133, 2018. 10.1080/17457300.2017.1341937 [DOI] [PubMed] [Google Scholar]

PERMALINK

The changing face of desktop video game monetisation: An exploration of exposure to loot boxes, pay to win, and cosmetic microtransactions in the most-played Steam games of 2010-2019

David Zendle

Rachel Meyer

Nick Ballou

Roles

Abstract

Introduction

Cosmetic microtransactions

Pay to win microtransactions

Loot boxes

The present research

Method

Ethics

Design

Measures

The presence of loot boxes

The presence of pay to win microtransactions

The presence of cosmetic microtransactions

Statistical analysis

Results

Changes in exposure to loot boxes

Changes in exposure to pay to win microtransactions

Changes in exposure to cosmetic microtransactions

Fig 1. Time series graph showing the percent of the sample playing games with each relevant feature during the period under test.

Discussion

Limitations

Conclusions

Data Availability

Funding Statement

References

Decision Letter 0

José C Perales

Roles

Author response to Decision Letter 0

Decision Letter 1

José C Perales

Roles

Author response to Decision Letter 1

Decision Letter 2

José C Perales

Roles

Acceptance letter

José C Perales

Roles

Associated Data

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases