The Data Told Two Different Stories. My Job Was to Figure Out Why.

Last year I took the plunge and finally started investing in myself again: I decided to go to grad school. What began as a "let me just do this so people will respect me more" has turned into a deep, passionate love for research and audience analysis, so you can only imagine my joy when we were given a real-life client to work with over the summer semester.

A premium childcare provider in the LA area that built its whole model around flexibility was losing enrollment to a new, free public alternative, and needed to understand how they were supposed to "compete with free."

While the initial problem seemed obvious, the real issue turned out to be a lot more interesting, and so did our research.

We didn't rely on one method — we used three

My team and I knew right off the bat this would not be a simple survey-type study. Choosing childcare is a deeply personal, multi-faceted decision, and we also knew that what parents report caring about isn't always what actually gets them to act.

So we ran a mixed-methods study: a 144-parent survey measuring what factors parents said mattered most, an experimental message test comparing three different value propositions (convenience, educational quality, and continuity of care), and eight in-depth qualitative interviews with parents actively making childcare decisions.

Each piece was designed to answer a different question. The survey told us what parents said they valued. The experiment told us what actually moved them in the moment. The interviews told us why.

Then the data seemed to contradict itself

The survey was clear: parents rated trust (6.48 out of 7), teacher quality (6.35), and educational quality (6.04) as the most important factors in a childcare decision. Flexibility ranked dead last.

But when we tested actual ad messaging, the opposite happened. The convenience-and-flexibility message outperformed every other message — highest likelihood of choosing premium care, largest effect size of the three conditions we tested (η² = .141).

Two data sets. Two different answers. If we'd stopped at either one, we'd have handed our client the wrong story with a straight face and a nice-looking deck.

This is where the interviews — and a decade of writing content for real audiences — actually mattered

Regression analysis gave us a clue: the same model that showed convenience mattering (β = .216, p = .035) also showed that believing quality was "worth paying for" was an even stronger predictor of choosing premium care (β = .307, p < .001). Both were true. They just weren't happening at the same moment.

I've spent years managing content calendars and community management for brands where the same audience needed a completely different message depending on where they were in the decision — a follower seeing a brand for the first time needs something different from someone about to convert. So when the survey and the experiment seemed to disagree, my instinct wasn't "one of these is wrong." It was: these are probably two different stages of the same journey wearing different outfits.

The interviews confirmed it. Parents told us, almost word for word, that convenience is what got them to look in the first place — but it was never what made them stay. One parent put it simply: it wasn't the easiest decision financially, but once she trusted the teachers, it felt like the right one. Convenience opens the door. Trust closes it.

Turning a contradiction into a framework

That reframe became the backbone of our recommendation: a four-stage "Parent Decision Journey" — capture attention with convenience, build trust with real teacher relationships and classroom transparency, justify the investment with concrete evidence of quality, and keep reinforcing that value after enrollment so families don't spend the next year second-guessing a decision they already made.

None of that comes from a single data point. It comes from being willing to sit with data that doesn't immediately agree with itself, instead of just reporting whichever number sounds cleaner in a slide.

Why this matters to me

I didn't come into this research background as a statistician, and I'm still not one. I came in as someone who's spent years figuring out what makes an audience actually do something — through content calendars, community management, and campaigns that lived or died on whether people engaged, not on whether the creative looked good in a deck. What this project reinforced is that those two skills were never as separate as I used to think. Reading a regression table is only half the job. The other half is knowing what a "contradiction" in the data usually means: not that something's wrong, but that you could be looking at two different moments in the same relationship.

That's the lens I want to keep bringing to research — not just what the numbers say, but what they mean for the actual person on the other end of the decision.