Inference

Loading...

Из курса от партнера University of Maryland, College Park

Framework for Data Collection and Analysis

243 оценки

University of Maryland, College Park

Framework for Data Collection and Analysis

243 оценки

Курс 1 из 7 — Specialization Survey Data Collection and Analytics

This course will provide you with an overview over existing data products and a good understanding of the data collection landscape. With the help of various examples you will learn how to identify which data sources likely matches your research question, how to turn your research question into measurable pieces, and how to think about an analysis plan. Furthermore this course will provide you with a general framework that allows you to not only understand each step required for a successful data collection and analysis, but also help you to identify errors associated with different data sources. You will learn some metrics to quantify each potential error, and thus you will have tools at hand to describe the quality of a data source. Finally we will introduce different large scale data collection efforts done by private industry and government agencies, and review the learned concepts through these examples. This course is suitable for beginners as well as those that know about one particular data source, but not others, and are looking for a general framework to evaluate data products.

Из урока

Quality Framework

In this module you will be introduced to a general framework that allows you to not only understand each step required for a successful data collection and analysis, but also helps you to identify errors associated with different data sources. You will learn some metrics to quantify each potential error, and thus you will have tools at hand to describe the quality of a data source.

Quality of Data5:01

Познакомьтесь с преподавателями

Frauke Kreuter, Ph.D.
Professor, Joint Program in Survey Methodology
Adjunct Research Professor, Institute for Social Research
Mariel Leonard
Lecturer
Joint Program in Survey Methodology

So when you think of inference, it is interesting.

There are really several steps that are important to keep in mind.

There are two types of inference in the survey setting,

one is the inference from a respondent to the characteristic of a respondent.

And one is the inference from the characteristic of the sample

to the characteristic of the population.

What do we mean by that?

Let's focus on the first part.

The respondent to the characteristic of the respondent.

Let's say you have an answer, very simple one.

For example, the answer to the question, how old are you?

You measure age, and you're interested in a larger concept,

which is aging process of the cells, of the mind,

experience, life experience in some form or what not.

So, you get the respondent's answer, but

the characteristic, aging characteristic that you're interested in,

is already an inference leap from that particular item that you're interested in.

And the answer to that item to the general concept that you're interested in.

So, that's one piece of inference that we are doing here, all right?

You infer from my answer to the question each.

Something about my experience.

My living conditions, my health, and the like.

Now, with all the answers to the individual responses and the inferred

characteristic with it, we can do some statistical computing, all right?

So, we form an average of all the respondents.

Answers and the implicit characteristic.

If you come from psychology, you might be used to having multiple measures

that form together a dimension, a psychology trait, a characteristic

on a measurement scale that of course would also be in this first part.

But then again, averaging that characteristic over all of the respondent

would be one form of statistical computing that gets you at

the characteristic of your sample, so you have values for everybody.

And average age of a classroom and average age of a population.

Careful of the population, only if you ask everybody.

If it is a sample, then at first you have the average age of the particular sample.

So, I take from all students of a university, a sample of students,

let's say its random.

Thousand students from a university that has 35,000 and I compute the age.

That would be the step under here.

Then there is an inference piece from

that characteristic of the sample to the characteristic of the population.

So if my sample average for

the university students who are, let's say, 21, then I can

do an inference to the entire population depending on how I drew my sample.

I might directly infer that that same value

is the characteristic of the population.

But in order to do that, you have to design your survey in a certain way

that these two inferential pieces do make sense.

And that's what we're gonna talk about when we look at the different

design decisions for a survey.

But keep in mind, the two parts where inference happens.

We will introduce one as sort of an aspect of measurement.

A characteristic of measurement, an inference about measurement.

That's the first part.

And the other one, that's a characteristic of representation.

Do the people that you gather information from represent and

allow inference to the entire population?

Ознакомьтесь с нашим каталогом

Присоединяйтесь бесплатно и получайте персонализированные рекомендации, обновления и предложения.

Coursera делает лучшее в мире образование доступным каждому, предлагая онлайн-курсы от ведущих университетов и организаций.

© Coursera Inc., 2018 Все права защищены.

Загрузить из App Store

Загрузить в Google Play