The answer to your question really depends upon your research question and the data involved. For example, is the 3/4 male outcome a true representation of the male/female distribution in the sampled population? If you are trying to make a population based statement, gender bias (if there is one) is part of the population and to "correct" for it would be a misrepresentation of the population as a whole. If the 3/4 ratio is incorrect, then the bias lies in the sampling strategy. An important question would then be -- why was the bias in the data collection? To correct for this bias, you would probably want to weigh the data sample to better reflect the "actual" ratio. There are other ways to address this, but like I started, the best method really depends upon the research question.

I liked Thomas's answer, except you need to be very careful in "correcting" the problem by weighing the samples. If the sample from the female population is biased, then the weighting will also increase the influence of this bias. So you really need to focus on why gender is skewed (or even if it is skewed).

Can you go back and take another sample with the main goal being to bring the sex ratio of surveys closer to the ratio in the studied population? Of course with this approach you will need both males and females because the previous survey effort could now bias the results in the this survey. You may also have trouble with not sampling the same individual twice.

Both Thomas and Timothy gave very good answers! The one question that is missing here is "how was the sample collected?" If it was a simple random draw of high school students then the data should be representative of the entire school (and I assume that is what you are trying to estimate to). If convenience sampling was used, then there is good reason to believe this is a biased sample, and, as the two responders noted, you may not be able to draw valid conclusions from your data (even by weighting). 

Simple answer: Run separate analyses for males and for females.

Problem with the simple answer is that we expect a sex ratio closer to 50-50. The bias that is skewing the sex ratio may also introduce bias into the data for each sex. So we return to Ariel's question: "how was the sample collected?" This includes every detail, from who interacted with the students, how questions were asked, where they were asked. These factors will interact with social norms, possibly resulting in unintended bias. As Thomas said, it may also be nothing more than the result of having surveyed a population that is naturally skewed male.

The other problem is that the question asked by Aparnna involved "gender based differences." If I analyze the data separately, then I am reduced to evaluating main effects by looking for non-overlapping confidence intervals. This will reduce the number of journals that will accept the manuscript -- pick your ideal journal and look to see what they require for reporting of statistical tests.

Why not use a stratified sampling technique?

You should use suitable sample design corresponding to your objective. The convenient sampling is not appropriate to collect data from school rather stratified sampling is suitable to collect data . Since convenient sampling is a subjective sampling and result is not efficient due to the bias estimate. otherwise you should be used Probability proportional to Size(PPS).