Choose the regression line before substituting

The source explicitly selects $y$ on $x$ for case (a). Substitute $x=14$ into the given model: $y=1.188(14)+3.042=19.674$. Reported here to 3 s.f.; the visible numerical working is blank.

Because $x$ is still prescribed explanatory, use the same $y=1.188x+3.042$ and solve $11=1.188x+3.042$. Thus $x=(11-3.042)/1.188=6.698653\ldots$. This authored blank completion is not the distinct $x$-on-$y$ fitted estimate.

The applicable fit is now $x$ on $y$. The source prints $14=0.736y-1.009$; completing its blank calculation gives $y=(14+1.009)/0.736=20.392663\ldots$.

The printed calculation is $x=0.736(11)-1.009=7.09$ to 3 s.f. Before rounding it is $7.087$. Here $y$ is the prescribed explanatory input.

Neither explanatory role is prescribed, so predicting $y$ from a supplied $x$ calls for the $y$-on-$x$ fit. Substitute $x=14$ to get $19.674$. This completes the blank source choice and calculation.

For a supplied $y$ and requested $x$, use the $x$-on-$y$ fit: $x=0.736(11)-1.009=7.087$. This completes the blank source choice and calculation; it differs from case (a)(ii) because the modelling roles differ.

With $y$ explanatory use $x=0.736y-1.009$, giving $x=0.736(30)-1.009=21.071$. The numerical completion is authored. The source explicitly notes that $30$ lies outside the observed $7\le y\le22$, so this is extrapolation. Check the range of the explanatory input $y$, not only the range of the resulting $x$. A high sample correlation does not establish the relationship outside that input range.