Many times researchers are working with free/open source software. Why they not go for commercial software which is more effective then opensource software.
I can think of several reasons (apart from price):
Commercial software usually comes with no source code, and no means to verify or check the algorithms therein;
Commercial software may be using proprietary data formats, rendering data files useless once the software becomes unavailable or obsolete;
The audience of your research may not have access to the same software, and thus may be unable to reproduce the results;
Open source software can be modified if necessary, e.g., you can add features that are needed for your research.
Also, I would take argument with the assertion that commercial software is "more effective". It may sometimes be true, but it is certainly not a general truth. By way of example, I use both Maple (commercial) and Maxima (free/open source) [disclaimer: I am one of Maxima's developers] for computer algebra. For some tasks (e.g., solving differential equations), Maple is more effective, but for other tasks (e.g., symbolic tensor manipulation), Maxima beats Maple hands down.
Certainly, when you are researching you are finding results, but how other researcher can replicate your results or validate your proposal y base code or base data is unavailable due non open source mechanism. Like V. T. Toth says, sometimes non open source software has high performance in comparison with the relative opensource software, but you should think in diferences where a team is working and where a community is working.
I think not all researchers uses open source application, because some times the researcher needs to process data using a known software. therefore, the situation determines the software type.