What about other options?

Let’s say you’re interested in this kind of non-experimental quantitative data, but you have reason to avoid a survey (or you’re looking to supplement one) – what other methods do we have for quantitative data collection? Let’s explore two more common ones: observations and secondary data.

Observations

Observing participants is yet another way we can get important research data. As the old saying goes, “actions speak louder than words,” and in some cases, actions are what you want to focus on. You might be asking yourself why we’re talking about observation in a chapter on quantitative data, that is how do we make numbers of these behaviors? Great question! We don’t always make them into numbers – so we will talk a bit about observation when we talk about qualitative data too – but there are quite a few creative ways to make this work.

Naturalistic observation

As a first step, let’s think of observations that take place in “natural” settings. Sometimes, a study can be so “natural” that participants don’t even know they’re being observed (and this can get tricky with the IRB, so be sure to have a conversation with an IRB representative as part of the planning of this kind of study). You might, for example, count the number of times you see a person with certain characteristics – how many adults pushing strollers do you see entering a grocery store, for example? Or maybe you’re looking for certain behaviors that you’ll rate – or code – for intensity or other traits in the people you’re observing. An interesting example of this is a naturalistic observation of childcare in public spaces (Amato, 1989). In this study, the researchers “stationed themselves in positions where they could observe pedestrian traffic” (pg. 984) in dozens of restaurants, parks, zoos, shopping centers, a football game, and a street festival in two different cities.

For each type of location, the observers were coding for different behaviors and situations. For example, in the recreational and commercial locations, like the shopping centers or football game, the researchers watched for children who appeared to be five years old or younger, then coded who was acting as the child’s “primary caregiver” – that is, “the person carrying the child, pushing the stroller or carriage, or holding the child’s hand” (p. 984). If only one adult was with the child, they were noted as the primary caregiver. The child and adult were observed for usually only one minute. In the restaurant settings, the behaviors being coded as “primary caregiver” were “setting up a high chair, assisting the child with his or her food, or caring for the child if she or he cried or fussed” (p. 984). In both cases, if two adults were sharing these tasks equally, or it couldn’t be determined who was primarily caring for the child, the case was not coded (about 5% of observations).

In each case that a primary caregiver was identified, the researchers coded the primary caregiver’s sex and ethnicity, the child’s estimated age and sex, the composition of the group (how many adults with the child and of what sexes), and the type of location.

What do you think of this observational scheme? Do you think the way this study was set up accurately captured the primary caregiver of these children in public settings? Would there be other aspects of the experience that you’d look for in this kind of observation? Whether you agree or disagree with the premise and set-up of this study, it is an impressive example of a lot of time spent in short, highly-detailed but natural observations – most of these adults and children probably had no idea of their involvement in research!

Structured observation

Natural observation have many strengths. For one, there is no chance of the Hawthorne effect (people changing their behavior due to being watched) – as long as the researcher isn’t noticed. This leads to high external validity in most cases. The observation is very real. However, there are times when we need to see what happens in very specific situations, or in situations that don’t usually take place in public. In this case, a structured observation may be in order. This would be treated much like any of the typical designs we’ve talked about so far in terms of sampling and recruitment. The participants would sign up for the study, consent to taking part, and be told as much as appropriate about the study (with appropriate plans for debrief if needed). Then, they’re put into a situation and watched (or, if not watched live, recorded). Let’s look at an example of childcare again, but this time, with a twist – observing parents before their baby was born. In the New Parents Project, a study of cohabiting and married first-time, different-gender parents in Ohio, the moms and dads were observed in the third trimester of their pregnancy. They were given a doll (a green sack-like object – it wasn’t a traditional babydoll) and asked to perform a series of “play” tasks with it according to a structured scenario. They were recorded while doing these tasks, and then their behaviors were coded for a variety of measures, such as coparenting quality and parenting qualities. Then later, then the real babies were about nine months old, the parents were again recorded doing the tasks with their actual child.  The researchers then looked at the similarities between the two observations; it turns out coparenting behaviors can show up even before there’s a real baby to parent (Altenburger et al., 2014)! In this case, the setting probably couldn’t have been more “fake,” but the results that were found seem very real. Sometimes a high level of control is needed so that the observations can be of very specific or rare behaviors. It all depends on what’s being studied and how the measurement of these variables is defined.

References

Amato, P. (1989). Who cares for children in public places? Naturalistic observation of male and female caretakers. Journal of Marriage and Family, 51(4), 981-990.

Altenburger, L. E., Schoppe-Sullivan, S. J., Lang, S. N., Bower, D. J., & Kamp Dush, C. (2014). Associations between prenatal coparenting behavior and observed coparenting behavior at 9 months postpartum. Journal of Family Psychology, 28(4), 495-504. 10.1037/fam0000012