Thursday, May 04, 2006

Sibling Effects on Exercise

What is the effect of having an older brother/sister in the house on a person’s decision to exercise? How does that effect change as the age gap between siblings changes?

It is hypothesized that older brothers are more prone to play sport, and will increase the probability that the respondent participates in sports. With an older brother, the supply of potential teammates increases, making it cheaper to participate in the activity because fellow players are easier to find. The respondents’ demand for athletics also increases because of peer effects; as respondents see those around them playing sports, they will be more likely to want to play as well. This argument hinges on the assumption that boys are more prone to play sports than girls. As a result, older sisters will likely have less of an effect or no effect on an individual’s decision to play a sport.
As the number of years between siblings ages increases, the demand for athletics will decrease because of fundamental differences in children’s growth rates. For younger children, there is less incentive to play with the much older children because of large skill differences that may make the activity less beneficial. Furthermore, the supply of siblings willing to play may decrease as their benefits decrease from playing with much younger and, therefore, more unskilled children.

The data set used is the National Longitudinal Study of Adolescent Health.
The main regression I am going to run regresses the age gap between the respondent and the next-oldest sibling with the gender of the sibling on the respondent’s decision to exercise in order to determine the effect on the respondent’s decision to exercise. Other variables will also be controlled for.
Y = b0 + b1(age gap) + b2(gender) + b3(age gap * gender)
Furthermore, their participation in group sports may be compared to their participation in individual sports.

Concerns and Limitations:
Certain individuals must be removed from the data set in order to make this regression as accurate as possible. For one, those respondents who have older siblings who are out of the household and therefore are not accounted for in the survey must be removed from the data set. This is because, for those respondents, the effects of their older siblings on their decisions to begin to play sports would not be accounted for and would distort the regression. As a result, there may not be very many results to examine for students on the upper end of the ages in the data set.


In your paper, be sure to very carefully describe why and how siblings might affect the decision to play sports. Think about what the precise effect of sisters and younger siblings are likely to be as well. This should help you more clearly understand your empirics.

Also, how do you plan to deal with cumulative sibling effects? I.e., what if I have 2 older brothers? How am I treated in your regression? Are you going to limit the sample to only 2 child families? Focus only on the effect of the closest in age sibling (controlling for the number of other older younger siblings)?
I think one thing to consider is that often times siblings with to distinguish themselves from each other. According to your hypothesis, identical twins should be the most likely to influence each other in their athletic habits since their skill levels are likely to be comprable and they are the same age. However, casual observation indicates that this might not be the case. While it is possible that one twin would want to get involved in athletics we often see the other sibling get involved in other activities. This effect might create some static in your results by adding an additional cost to conforming to the behaviour of a sibling. Also, it is unclear in your proposal whether you are including the siblings decision whether or not to exercise in your regression. I definitly think this variable will have an effect on a siblings decision to exercise and should therefore be included.
You stated in your assumptions that boys play more sports than girls. This does not have to be an assumption. There is probably data somewhere citing participation numbers for boys and girls, giving you the chance to strengthen your claims and hypotheses.

Also, I think you may want to consider different transformations for the age gap. Your reasons for stating that sibling effects might decrease as the age gap increases are valid, but maybe as the gap gets bigger (6 or more years) the younger sibling could react differently to his older brother. At this point, the older brother begins to enter role model status and slowly departs from rival status. Of course, this does not affect your other hypothesis that the supply may decrease. The way I describe this affect would be equal to a power transformation, achieved by adding the variable agegap^2.
I think this is a great topic, and one that I have a personal connection to with 2 brothers. My one concern is that I think a lot of what you're testing has an endogenous relationship. The motivation and desire to exercise is often common in a family, and this is not necessarily because of sibling influence. My brothers were very much into sports, but this was not why I got into sports. I think it's more genetic for my family (an endogenous effect)--both sides of my family are extremely athletic. I wonder if there's a way to control for this sort of thing.
I agree with Will, and think you need to make sure that your regression reflects your hypothesis. In other words, even if you find that children with an older brother are more likely to play sports, it may not be correct to attribute this result to the brother. Older brothers may simply play sports because the family places a high value on sports, and this will also lead the younger sibling to play a sport as well.
Post a Comment

Subscribe to Post Comments [Atom]

<< Home

This page is powered by Blogger. Isn't yours?

Subscribe to Posts [Atom]