Their method cannot account for all of the variables that affect well being because no one fully understands what all those variables are, much less having the ability to account for them in a study like this.
Per your links they accounted for just 6 variables, not nearly enough to make the stated conclusion with any confidence:
> In addressing our second research question, our aim was to approach the independent contributions of internet access and use on well-being. To that end, we adjusted for plausible (and available) covariates in our models that might otherwise mask or bias any independent contributions of internet access on well-being or create spurious associations. We chose six variables to represent such potentially confounding factors that have been previously considered important in the literature on well-being (): the respondent’s income (e.g., ), educational (), work (), and relationship statuses (), their ability to meet basic needs for food and shelter, and whether or not they reported having health problems ().
Per your links they accounted for just 6 variables, not nearly enough to make the stated conclusion with any confidence:
> In addressing our second research question, our aim was to approach the independent contributions of internet access and use on well-being. To that end, we adjusted for plausible (and available) covariates in our models that might otherwise mask or bias any independent contributions of internet access on well-being or create spurious associations. We chose six variables to represent such potentially confounding factors that have been previously considered important in the literature on well-being (): the respondent’s income (e.g., ), educational (), work (), and relationship statuses (), their ability to meet basic needs for food and shelter, and whether or not they reported having health problems ().