sleep and behavior on average teenagers.

[kailamaria]

Hi, I am a freshman and i am doing a science prodject on. How the amount of sleep the average teenager gets affects their behavior inside and outside of school?

Could you give me information about a good servey to colect my data.

Thank you,
Kaila.

[aelin]

Hi Kaila,

The question that you're asking is very relevant today and definitely has significant implications on behavior. Eg, how will I behave if I get more sleep (which I probably need =/). Sorry for the lengthy response, if you just read a couple of sentences from each point (in bold), you should be fine. There are an excess of details in here though because I enjoy talking/writing, as you will soon find out.

Surveys of this type are generally very tricky to construct, but with the right effort and focus, they reap huge benefits. I'll break it up into 4 main categories: picking the right sample, constructing a survey, receiving data, and interpreting the data (this is the way I generally see/do it, don't actually quote me in a real statistics course, though this does cover the vast majority of sampling issues. You can find more details here, but its rather lengthy: http://en.wikipedia.org/wiki/Statistical_survey).

1. Choosing the right sample - what I mean by this is, who do you need to survey to sufficiently answer your question? Who is the 'average teenager'? This ties in very closely with the issue of representativeness; if a sample is representative of the population (eg a particular group of teenagers is representative of all teenagers), then this makes it more likely that our sample is a relatively accurate reflection of the population. So in your case, it would be rather difficult to find a group of teenagers that would qualify as representative. In that scenario, you may find it easier to narrow the scope of your question to 'the average teenager at my school' since it will probably be easiest to sample teenagers at your school.

From there, you can ask the school for a roster and design some kind of random number generator or algorithm to randomly select however many students you want to sample (generally, statisticians use around 10% of the whole population, but this may be difficult depending on the size of your school). Any number greater than 30 should be sufficient though. Ideally, the random sampling prevents or minimizes the potential impact of bias on the sample responses.

2. Constructing the survey - obviously, one key thing that you need to survey will be the average number of hours of sleep each participant in your study receives. The construction of the survey itself, generally in wording, should be kept simple. Something along the lines of "On average, how many hours of sleep do you get per night?" should be fine. Make sure that the wording is clear though; if necessary to the goal of your study, differentiate between school nights and weekend nights. Ask exactly what you want to know.

Secondly, you want to think about how you want to construct the responses. Do you want to supply the participant with checkboxes of ranges of hours of sleep, or do you want them to simply write down a number? My preference is for the former, as I believe that it helps standardize responses so that participants generally won't write silly things like 4124 hours. But again, this depends on what you feel like knowing. Ask what you want to know.

3. Receiving the data - Ok, so I just dealt with the survey itself, and I've deliberately kept it separate from 'receiving the data,' by which I mean collecting the behavioral data of the participants. Along with a survey, I suggest that you implement some kind of journal that participants use to keep track of their actions. In a sec, I'll explain why, depending on how I'm interpreting your study, I've separated this from the survey. First though, I support the use of the journal as it is highly unlikely that you can monitor each participant to see what they are doing.

That being said, I've kept this separate from the survey for a few reasons. First, many behaviors that a researcher like yourself might notice are not readily able to be recollected by a participant. So, while completing a survey, a participant may have no idea how many times he/she did X behavior, depending on the behavior. Second, behaviors may be interpreted differently by a researcher and a participant, also depending on the behavior. Now, you may have noticed that I added the caveat of 'depending on the behavior.' Clearly, some behaviors are more noticeable/distinguishable than others, so depending on what behaviors you are trying to observe, a survey might work just as well or even better than a journal, especially since participants may not be super excited to have to write stuff down. Eg, if you are trying to measure how many times a person took a nap per day, a survey would probably work fine; whereas if you are trying to measure how many times a person raised his/her voice, handing them a journal and asking them to record each instance might work better. Again, I leave this up to your discretion. If the survey seems like the better option, see (2) above and again, ask exactly what you want to know.

4. Interpreting the data- However, if you do choose to go with the journal route, be sure to clearly define each behavior down on paper BEFOREHAND. Make it clear what constitutes and example of X behavior and what does not. This will save a lot of potential confusion later on. As for the statistics, you can leave that for later, feel free to ask again if you need help with that.

Cheers! Hope all goes well,
Aaron Lin

[deleted-42343]

We have some resources that talk about how to do a survey:

https://www.sciencebuddies.org/science- ... rvey.shtml
https://www.sciencebuddies.org/science- ... ants.shtml

You might consider something easier to measure than a journal. The generic test is always testing reaction time of the participants. There are probably many other tests you could do, too. As for measuring their actual behavior, that could be very difficult. You can't just ask them if they feel grouchy when they have had less sleep (many won't even realize what they are missing!). I would try to read some research papers on sleep to better understand how the experts conduct these experiments. Unfortunately, you don't have a sleep lab (a controlled environment), where you can do lots of tests on people and force them to stay up later or sleep more (plus, these people are normally paid for their time). But doing some background research might give you some simple experiments you could do with people.