Looking for help on how to prove the following inequality:
$c\,(c-1-x)\,{}_2F_1(1,1\,;\hspace{1pt}c\,;x) \;\,\le\;\, (c-1)\,(c-x)\,{}_2F_1(1,1\,;\hspace{1pt}c+1\,;\,x), \;\;\; 0\le x<1,\; \, c>1 $
I have tried using Gauss contiguous relations, and linear transformation formulae from A&S and G&R. I have looked at the difference of the coefficients in the expansions.
Alternatively:
$$\frac{c\,(c-1-x)}{(c-1)\,(c-x)}\,{}_2F_1(1,1\,;\hspace{1pt}c\,;\,x)\;\;\le\;\; {}_2F_1(1,1\,;\hspace{1pt}c+1\,;x) \;\;\le\;\; {}_2F_1(1,1\,;\hspace{1pt}c\,;\,x),\\\quad\quad\quad\quad \qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad\qquad \;\; 0\le x<1,\; c>1 $$
Also apparently true for $-1<c<1$. This arose in my research trying to prove convexity of the bias of $R^2$ in multiple linear regression.
Thanks.
I figured it out. Using the integral representation, the right side minus the left side of the first inequality above gives:
$\int_0^1 \frac{(1-t)^{c-2}((c+x)t+1)}{1-tx} dt$
which is positive for $0\le x<1$ and $c>-1$.