Protocols for estimating age of teens through observations? | NAAEE

The hub for environmental education professional development

Protocols for estimating age of teens through observations?

Hi everyone,

I have a colleague working on a data collection protocol, and we have a question I'm hoping to crowd source through all of you. He's interested in observing how park visitors interact with wild horses on Cape Lookout National Seashore. They'll train field techs in an observation protocol, and one of their tasks is estimating age of the visitors. In the past, they've used the following age categories (previously used before in similar studies):

Child = Children from infancy to 12 years of age as children.
Teen = Code adolescents from 13 to 20 years of age as teenagers.
Adult = Code people from 21 to 59 years of age as adults.
Senior = Code people 60 years of age and older as seniors.

They are interested in tracking how visitor behavior and activity, as moderated by distance, affect horses' behavior and activity budget (kind of "forced energy expenditure" from the horse's perspective). The PI was curious as to whether these categories are appropriate for environmental behaviors of children. My take is that most of the EE lit focuses on intended rather than observed behaviors, but that the 12-20 age range does seem to be pretty broad in terms of differences between late elementary/early middle and high school students in terms of environmental engagement and intended behaviors. I thought he might try splitting the group into 12-15 and 15-20 and seeing if there's a difference (and aggregating if it doesn't work out).

So - questions for the group:
1. Any more thoughts on this age grouping from your perspective?
2. If we wanted to partition kids ages, any protocols out there for helping field techs eye ball ages of kids without actually talking to them?

Thanks for any and all resources!


My first question would be how far away are the observations to estimate age taking place? Interaction with others would be helpful to give a better indication of the age of the youth being observed.
Plus, how old are the techs doing the observations? that will skew, unconsciously, their thoughts on ages of youth.
We all have those students we work with that we go, WHAT? you are how old?! because they look older or younger... and our own thought puts them at a completely different age, especially when we aren't interacting with them.
So I would be hesitant to break up youth into too many categories. Perhaps if you could observe that they were driving, that would make it easier to estimate 16-20.

Hi Kathryn. It looks like the research will be primarily on variables of visitor behavior, distance, horse behavior and activity budget. If age is not specifically included in the research question, I would say your colleague's age categories seem appropriate for the study. If the first round of data reveals some kind of pattern with age, then he/she may want to consider subcategories like adolescence (13-17) and young adult (18-22). Making sure there is a large sample size of observations and having two coders present and establishing inter-rater reliability on age codes, or all codes for that matter, would add validity to the data collection protocol and interpretation. Observational studies are notorious for requiring more resources, but it is worth the effort. Good luck to your colleague on this important work.

As an education professional, not necessarily a researcher, what struck me was the broad range defining "children." When considering brain development, I might suggest considering additional breakdowns to reflect that:

Early Childhood = birth to 8 years of age
Children = 9 to 14 years of age
Young Adults = 15 to 20 (or even later - brain development is still happening, especially in males through mid-20s)
Adult = Code people from 21 to 59 years of age as adults.
Senior = Code people 60 years of age and older as seniors.

Thanks everyone!

Betsy, Hillary has it about right, and this is how I kind of thought we should proceed, too. I just wanted to crowdsource to make sure I wasn't missing anything. My thinking for maybe splitting was along the lines of Victoria's (and the potential to ask a side RQ), but I'm guessing for the reasons you've all brought up, the categories are probably ok. I'm not in charge of this particular part of the project, but I think the protocol does require multiple observers and inter-coder reliability.

Thanks again - and if anyone thinks of anything else, please do pass along!