Ok, well first of all...
You are dealing with different samples, a random selection of conservatives is not the same as a random selection of liberals.
There interests are not the same.
That is of most primacy.
Second of all, there is testing methodology and this is probably most important. What does it mean to be informed or misinformed on current events and which events? Are we asking liberals if they are aware of weekend anti-Globalist French protests to the same degree we are asking conservatives where the EU is headquarted? If we ask were is the EU headquarted, why not ask where NATO is headquarted? Who is likely to know the answer to which question?
Third of all is viewership. What does it mean to watch Fox News, some shows do very little current events at all and focus solely on an issue or two which they go in to detail. The average fox viewer watches for far longer than those other platforms. Does an NPR listener count as news or they listening to car talk only? The test likely doesn't account for these disparities, they never do (ideologically bent studies) they never take all variables into account sufficiently.
From one variable you necessarily conclude one conclusion, the one that you want to from the data. You don't sufficiently isolate that variable from the many, many other variables... doing so would one, likely not produce the same result you are looking for and two, be way more difficult than you realise. Any scientist worth his salt knows that it is much, much harder to know what you think you know.