The professor starts promptly at 4:10 on a cool October day:
With pooled cross-sectional data, we are not measuring the same unit, person, state, at two or more points in time, what we are doing is we’re taking a random sample at a time period, and then taking a fresh random sample from the same population at a second time period, and so on. So you are pooling together two or more random samples...
John sat attentively, pen in hand, paper out, and laptop humming. While listening to the professor, he jots down “pooled cross-sectional = two random samples at different times” onto his notepad.
...Cross sectional data allows you to ask certain kinds of causal questions. It is a little bit limited, so that is why we are going to spend next week and the following week looking at panel data and it really opens up a whole different set of questions we can ask because now we are measuring panel data at the same unit at two or more points in time! Does that make sense?
‘No,’ John thinks. He looks at what he just wrote down: “Cross-sectional = limited causal questions” into his notepad. That doesn’t seem right.
The professor was onto something else already. John heard the word “Heteroskedas... something.” He thought about raising his hand, but was not sure how to phrase the question. “What’s a Heteroskedas... [mumble, mumble]” was way too mediocre, he couldn’t ask that and reveal his total ignorance. After all, they may have spent the entire recitation reviewing that term. ‘If I ask that question, I’ll show my cards and everyone will know!’
Psst, his friend sitting to his left whispers, “What did he say about cross-sectional data?”
John shrugs. “I think it is limited.”
Buzz buzz. His phone vibrates on the table. John looks at it. His friend Mario texted him: “Hey, are you going to the Halloween Boat Party? How do you get tickets again?”
‘Oh, yeah,’ John thought, ‘The boat party. That looks pretty neat. I like boats, but something about being in my thin, tight Superman costume in the middle of the Hudson at the end of October doesn’t feel right.’
...let’s go through one example here. One starting hypothesis is there may be one single factor that is common to all seven variables or maybe it is multi-dimensional. The question is are these seven individual variables manifestations of...
Right below “Cross-sectional = limited causal questions”, John scribbles “starting hypothesis is a single factor,” even though he already knows that comment makes no sense outside of any context.
John checks his laptop. Looks like he got a few notifications on Facebook. Oh, looks like Ali liked a picture:
'Hahaha, this is great.' Share. He is stuck now. Maybe one more scroll to see what else is going on.
'How Trump made a Seeple? Meh'
'Haha, look at the little cameras. That’s not accurate.' Share.
...Creating an index out of these three ingredients would be a better, more valid, more comprehensive measure than just choosing one variable, like the answer to the survey question, “how important is religion on guiding your life?”...
John picks up his phone and starts typing: “Yeah, I’m going to the Boat thing. Prob be a bear, or Neo, something warm. Tickets are going to be sold in-person some night. Heard it is going to be PACKED”
...This index will give us a richer operational measure of this higher order construct, “religiosity,” than any single variable. So, in some ways, you are not only avoiding multicollinearity problems, it is more satisfying even theoretically...
The gradually rising noises from the hall signal that the end of class is approaching. Zippers start sliding and laptops are shut. Over the rustle of people packing up, the professor continues...
Don’t forget the midterm is next week. We will be holding extra office hours this week and recitation will be more of a review session...
On the way out, John bumps into his friend, Mengyao. She asks, “What did you think of the lecture?”
“Yeah, the professor is so engaging. I’m glad to finally be learning something concrete and applicable. The professor really brings the class alive, you know?”
“Yeah, I know what you mean!” She says, “Hey, are you going to the Halloween Boat Party?”