"The hypothesis says that an important positive relationship between YouTube and the AEC learning process was assisted by the data collected and the analysis. The result stated that a unit enlargement in YouTube increased by 55% in the AEC learning process because student uses YouTube very often to learn"

Let's simplify the hypotheses in notational form:

Ho: AECscore(YouTube) <= AECscore(No YouTube)

Ha: AECscore(YouTube) > AECscore(No YouTube)

Do you agree with this formulation? If so, note how you are trying to use data to support your claim -- the alternative hypothesis. Also note that the null hypothesis contains the equal sign, and that the pair of hypotheses are both mutually exclusive (must be one or the other) and collectively exhaustive (the pair comprise all possibilities).

"The threat of the study is based on the questionnaire in my opinion because 52% of them were rejected and weren't included in the analysis because of discrepancies in the responses and many of the questions were left unanswered."

Good point! Deciding how to treat non-responses is a rabbit hole within the world of measurement & assessment, but you're right to point out the threat to both internal & external validity it poses. Did the article discuss details on whether participants skipped particular questions? Or perhaps the survey was too lengthy??