1) You are incorrect. They do publish negative reviews. Michael Fremer, in particular, is not bashful about publishing a negative review.
2) Reviewers tend strongly to want to review components which are of personal interest to them, and which they suspect they will like. Sometimes a reviewer who owns, or who has reviewed favorably in the past, a particular component will be assigned the review of the next version of, or the successor to, that component. There are organic gravitational pulls which lead reviewers to review components they naturally are likely to enjoy. This naturally and legitimately drives up significantly the number of positive reviews, and the ratio of positive reviews to negative reviews.
3) Once in a while a reviewer will audition a component which has been sent in for review, give the manufacturer a heads-up that a review is not likely to be favorable, and the manufacturer withdraws the component from consideration. This also increases the number of positive reviews, and the ratio of positive reviews to negative reviews.