If you use the remember picture option of tikzpicture the picture origin is written to the auxiliary file (.aux). The used unit is scaled points sp which is 65536sp = 1pt = 1/72.27inch. The origin is IIRC the lower left page corner. You can also save any content in a box register and write its size into the auxiliary file or another one.
Even better is the zref-abspos package which allows you to get the absolute position of any point, also in sp. You need to place a \zsavepos{<name>} at these coordinates and then can get the coordinates after one compilation using \zposx{<name>} and \zposy{<name>}. The manual states that it is only good for relative positions because there is no official reference origin, but in all my tests it was also the lower left page corner.
The following code should demonstrate this:
\documentclass{article}
\usepackage{zref-user}
\usepackage{zref-abspos}
\usepackage{tikz}
\usetikzlibrary{calc}
\newcommand{\checkBox}[1]{%
\begin{tikzpicture}[color=black, line width=0.2mm]
\draw (0,0)
node {\zsavepos{#1-ll}}
rectangle (5mm,5mm)
node {\zsavepos{#1-ur}};
\end{tikzpicture}%
\write\mywrite{#1: \zposx{#1-ll}, \zposy{#1-ll}, \zposx{#1-ur}, \zposy{#1-ur}}%
}
\newwrite\mywrite
\openout\mywrite=\jobname.pos\relax
\begin{document}
Question:\\
\checkBox{Q1A}~Answer A\\
\checkBox{Q1B}~Answer B\\
\checkBox{Q1C}~Answer C\\
\checkBox{Q1D}~Answer D\\
\end{document}
It writes a .pos file which holds the positions like this:
Q1A: 9017948, 41890363, 9950287, 42822702
Q1B: 9017948, 40455628, 9950287, 41387967
Q1C: 9017948, 39020893, 9950287, 39953232
Q1D: 9017948, 37586158, 9950287, 38518497
The format can also be changed to get the values in pt or another unit, or the origin plus size instead of the absolute position of the lower left and upper right corner like it is now.
For mm output replace the \write content with:
\write\mywrite{#1: (%
\dimtomm{\zposx{#1-ll}sp},
\dimtomm{\zposy{#1-ll}sp}) + (%
\dimtomm{\zposx{#1-ur}sp-\zposx{#1-ll}sp},
\dimtomm{\zposy{#1-ur}sp-\zposy{#1-ll}sp})%
}%
and add the following macro:
\makeatletter
\newcommand\dimtomm[1]{%
\strip@pt\dimexpr 0.351459804\dimexpr#1\relax\relax mm%
}
\makeatother
Then you get:
Q1A: (48.36134mm, 224.64914mm) + (4.99992mm, 4.99992mm)
Q1B: (48.36134mm, 216.95496mm) + (4.99992mm, 4.99992mm)
Q1C: (48.36134mm, 209.26077mm) + (4.99992mm, 4.99992mm)
Q1D: (48.36134mm, 201.5666mm) + (4.99992mm, 4.99992mm)
If you mind the shown precision and like to round the numbers instead have a look at \pgfmathprintnumber of PGF/TikZ. However, if you parse these files anyway you can do it then as well.
If you are using DVI mode (i.e. latex not pdflatex), the tikzpicture is created using PostScript code executed by dvips after the TeX processing. In this case both nodes are not yet placed and get the same coordinates which should be the origin of the picture. Instead I would only save one point and measure the size of the picture to get the other one.
Note that I added some PGF/TiKZ code to make sure that \zsavepos doesn't influence the size of the picture.
\documentclass[a4paper]{article}
\usepackage{zref-user}
\usepackage{zref-abspos}
\usepackage{tikz}
\usetikzlibrary{calc}
\newsavebox\mybox
\newcommand{\checkBox}[1]{%
\begingroup
\sbox\mybox{%
\begin{tikzpicture}[color=black, line width=0.2mm]
% Pace the save pos "whatsit" without influencing the bounding box / size of the picture
\begin{pgfinterruptboundingbox}
\node at (0,0) {\zsavepos{#1-ll}};
\end{pgfinterruptboundingbox}
\draw (0,0) rectangle (5mm,5mm);
\end{tikzpicture}%
}%
\usebox\mybox
\immediate\write\mywrite{#1: (%
\dimtomm{\zposx{#1-ll}sp},
\dimtomm{\zposy{#1-ll}sp}) + (%
\dimtomm{\wd\mybox},
\dimtomm{\ht\mybox+\dp\mybox})%
}%
\endgroup
}
\makeatletter
\newcommand\dimtomm[1]{%
\strip@pt\dimexpr 0.351459804\dimexpr#1\relax\relax mm%
}
\makeatother
\newwrite\mywrite
\immediate\openout\mywrite=\jobname.pos\relax
\begin{document}
Question:\\
\checkBox{Q1A}~Answer A\\
\checkBox{Q1B}~Answer B\\
\checkBox{Q1C}~Answer C\\
\checkBox{Q1D}~Answer D\\
\end{document}
Without having this evaluated fully so far, it looks to me like a (known) deficiency in the LaTeX2e footnote handling if floats are part of the picture. The problem is that there is enough free space in the first column for LaTeX to read on until after the second section and the float flollowing it (without finishing the first column).
Now when LaTeX looks at this float it find that there is not enough space to put it into the first column. So it puts it on the deferlist and puts the text back to main vertical list for reprocessing. And there is the catch: The footnotes ended up a single box by that process (well here is only one, but assume you have several ones from several places on the page) Now these are reinserted as well but at the end of the accumulated text not in their original place. LaTeX makes the assumption that all the material it has seen will have space on the current page and will end there. This is normally true but not in your specific scenario.
In your case the the section 2 heading plus a line would fit into the first column, however only 1 line! But LaTeX prohibits breaking after the first text line after a section. So normally LaTeX would find out that it is not possible to put section2 + 2 lines into the first column and therefore would break before the section.
However, before LaTeX can detect this it sees your figure (directly after the heading) and therefore tries to place that figure and that moves the "thanks" after the section 2 heading. Only then it goes back and tried to typeset the first column and the section still doesn't fit with two lines after it, so it break before it but now your "thanks" are moved.
You can verify this by placing a \pagebreak into the first line after the section2. then this moves into the first column.
Alternatively, move the placement of the figure a bit after the heading. It is in fact not a good idea to have it directly after \section because of the special processing that the heading tries to do to the following paragraph.
Now this "is" a deficiency, but the 2e algorithm is so delicate that modifications here are not advisable.
Best Answer
You can raise the text using
\raisebox{<amount>}[<set height>][<set depth>]{<content>}and also change its official height and depth. The original dimensions are available as\height,\depth,\totalheight(height+depth) and\width.So
\raisebox{5cm}[\height][\depth]{<some content>}should move it 5cm up, but keep the original dimensions as they are, making LaTeX think the content didn't change.