Author(s) This pre-registration is currently anonymous to enable blind peer-review. It has 2 authors.
Pre-registered on 06/07/2022 12:45 PM (PT)
1) Have any data been collected for this study already? No, no data have been collected for this study yet.
2) What's the main question being asked or hypothesis being tested in this study? Brief Background: The popularity of watching gambling online, via streaming platforms such as Twitch and YouTube, has proliferated in recent years, but its prevalence and psychological effects among people who gamble are unknown. For example, it is important to understand whether regular gamblers watch gambling streams as a means of overcoming urges to gamble, and/or whether watching gambling streams actually increases urges to gamble. The present study is a large-scale follow-up of an exploratory study. We will compare various self-report measures between gamblers who watch versus do not watch gambling online.
i) Gamblers who watch gambling streams on streaming platforms such as Twitch or YouTube, will score higher on the PGSI, boredom proneness scale, and state gambling urge scale, compared to those who do not watch gambling streams.
ii) Among viewers of gambling streams, PGSI, boredom proneness, and gambling urge will all correlate with increased cravings from watching streams, whereas using streams to decrease cravings will not be significantly correlated with these markers of gambling harm.
iii) Problem gambling severity will also be associated with age (younger) and gender (man). We will test whether the above relationships hold in men and women separately.
3) Describe the key dependent variable(s) specifying how they will be measured. i) Problem gambling severity score measured using the 9-item PGSI. Scoring: Likert scale from 0 (Never) to 4 (Almost Always). Higher severity problem gambling corresponds to higher scores.
ii) Boredom proneness will be measured using the 9-item boredom proneness scale. Scoring: Likert scale from 1 (strongly disagree) to 7 (strongly agree). Higher boredom proneness corresponds to higher scores.
iii) Gambling urge will be measured using the 7-item state gambling urge scale. Scoring: Likert scale from 1 (strongly disagree) to 7 (strongly agree). Gambling urge corresponds to higher scores.
iv) Increased craving from watching streams will be measured using a 2-item scale. Scoring: Likert scale from 1 (strongly disagree) to 7 (strongly agree).
v) Using streams to decrease cravings will be measured using a 2-item scale. Scoring: Likert scale from 1 (strongly disagree) to 7 (strongly agree).
vi) Twitch watching will be measured using 1-item. Scoring: Likert scale from 0 (Not at all in the past 12 months) to 8 (4 or more times a week). Twitch viewers correspond to scores >1 and non-viewers correspond to scores <1. People who score 1 will not be included.
4) How many and which conditions will participants be assigned to? We don't have any conditions. Our predictors are described in the dependent variable and analyses section.
5) Specify exactly which analyses you will conduct to examine the main question/hypothesis. i) We will conduct separate t-tests comparing PGSI, boredom proneness, and state gambling urge between Twitch viewers and non-viewers.
ii) We will use Spearman's rank correlations to examine the associations between the 'increased cravings from streams' scale, and the 'using streams to decrease urges' scale, against PGSI, boredom proneness, and state gambling urge
iii) We will compare the correlation coefficient between PGSI and 'increased cravings from streams' with the coefficient between PGSI and 'using streams to decrease urges.' We will do the same for boredom proneness and state gambling urge (replacing PGSI with these variables).
iv) We will use Spearman's rank correlations to examine the associations between gambling urge, boredom proneness, and problem gambling
v) We will use Spearman's rank correlations to examine the associations between age and gender and problem gambling. We will re-run analyses 1 and 2 among men, and also women if our sample size permits.
6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations. No outliers will be excluded. Participants will be excluded based on the following:
- If they fail at least one of two attention checks: In two of our scales, we added: "In order to check the reliability of your responses, please select "Neither Agree Nor Disagree" as your answer to this question". Participants will be excluded if they do not select "Neither Agree Nor Disagree"
- If they completed the survey in under 4 minutes and 50 seconds: Brysbaert (2019 J Mem Lang) establishes the average speed of silent reading of online surveys as M = 238 words per minute (SD 51). Taking a speed 3 standard deviations above the mean, 391 words per minute, as implausibly fast, and based upon our main survey holding 1,888 words, we impose a threshold of 4 mins 50 seconds (see also cleaning guidelines from Buchanan & Scofield 2018 Behav Res Methods).
- If they are not current gamblers. We asked, "In the past 3 months, how would you describe your gambling activity? If they answer, "I do not gamble at all", they will be excluded. Note: We ask this in our pre-screen and present this a second time to make sure that our sample consists of gamblers.
7) How many observations will be collected or what will determine sample size? No need to justify decision, but be precise about exactly how the number will be determined. 1000. The completion point will be determined by Prolific.
We aim to have 800 participants in our main analyses. Based on our findings from a pilot study, we assume 300 will be stream viewers. Assuming 300 stream viewers (versus 500 non-stream viewers), we will have at least 90% power to detect an effect size of d=.24 or larger at an alpha level of 0.05.
8) Anything else you would like to pre-register? (e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?) To further test the robustness of our findings, we will run secondary analyses to test whether differences between Twitch viewers versus non-viewers on the PGSI, boredom proneness scale, and gambling urges could be due to differences in age (younger age ~ gambling streams) and gender (men ~ gambling streams). First, we will conduct separate multiple regressions to see if viewership (Twitch viewers vs non-viewers) predicts PGSI, boredom proneness, and gambling urges even when controlling for demographic covariables. Second, we will identify a third group who report exclusively watching non-gambling content on the same streaming platforms, because this subgroup is likely to be similar, demographically, to the Twitch viewers. We will run separate ANOVAs with the three groups (Twitch viewers, non-viewers, non-gambling viewers) predicting PGSI, boredom proneness, and gambling urge. Post hoc tests (Tukey's) will be used to compare specific groups.