The "chain rule with integration" is "substitution"- but you are doing it wrong.
If you had [itex]\int (ax- b)^n dx[/itex] then you could let u= ax- b so that du= adx and dx= du/a. Replacing ax- b by u and dx by du/a, the integral becomes [itex]\int u^n (du/a)[/itex] and because a is a constant we can take it out and get [itex]\frac{1}{a}\int u^n du[/tex]. <br />
<br />
But if we have [itex]\int (cos(x)- b)^n dx[/itex] and try to do the same thing, we run into a problem. Letting u= cos(x)- b, we have du= -sin(x)dx so that dx= -du/cos(x). Putting those into the integral, [itex]\int u^n (du/cos(x))[/itex]. But now we cannot take "cos(x)" outside the integral because it is NOT a constant! And we cannot integrate cos(x) with respect to u.<br />
<br />
The "chain rule" with <b>differentiation</b> says that if f(x)= g(u(x)) for some differentiable function u, then df/dx= (dg/du)(du/dx). That is, we multiply together the derivatives of g with respect to u and u with respect to x. But doing the "opposite", substituting in the integral for u(x), [itex]\int g(u(x))dx[/itex] we cannot just do the integration with respect to u of g and then multiply by the integral of g. We have to have something line [itex]\int g(u(x))u'(x)dx[/itex]. u'dx= du so the integral becomes [itex]\int g(u)du[/itex]. But the derivative, u', has to <b>already</b> be in the integral. We cannot just multiply as we can with differentiation.[/itex]