A win is a win but it wasn't a crushing defeat. Dukakis lost by the skin of his teeth, and could have easily won if he would've ran a decent campaign. Is it wrong to learn from the past? Of course not. If Dukakis would've ran a better campaign and gotten a few percentage points, the electoral vote would've reversed, because few of H.W.'s defeats were crushing.
It DOES matter that the election was close, you can't pretend that it doesn't. It undermines the presidents credibility and makes it apparent that he isn't going to stay in power much longer.
Interesting theory, but highly flawed. Your assumption is based on static conditions.
The victory was narrow, so the likelihood of re-election is small? While it sounds good on paper time and history waits for no election. George W. Bush lost the popular vote, so applying your logic he would naturally not be re-elected in 2004.
But wait! First there is September 11, 2001, the War on Terror and the invasion of Iraq. All of these were significant factors in the Republicans retaining control of Congress in 2002 and Bush successfully defeating Kerry. All three events led to a resurgence of nationalism that generated a corresponding rally around the President effect. Accordingly, the credibility gap, which more accurately should be described as political capital, was erased.
Approach it from the reverse perspective. Bush wins in 2004 and enjoys a Republican-led Congress. He states that having won both the Electoral College and the popular vote he now has considerable political capital.
In the spring of 2005 he proceeds to spend that political capital in his efforts to reform Social Security. Based on your logic, he should have succeeded. However, even though he did a cross-country tour and gave numerous speeches, a Republican-led Congress wouldn't have take up his proposals.
As I tell my students, while it may look good on paper, sometimes it just doesn't quite jibe in the real world.