Template Chapter (Text)

We would have big sections like "text data" and under the section, we would have the smaller specific chapters (e.g., dictionary-based approaches). The titles should focus on the types of data. For example, "How to collect and analyze text data using computational methods"?

Introduction

A fun description of the data that is available out in the wild for the social scientists. For example, student reflections. Make sure to discuss a few common data types (interview transcripts) and bridge that gap.

Data

How does the data look?

It definitely needs to have a text column. It can also have other variables entered (maybe add a screenshot).

How can you get this data?

From your classrooms, out in the wild (Twitter). Even scraping web pages, etc.

Getting Started with Research

Research Questions

Now that you know how and where you can get text data, and how you can create a data file, now let's talk about what kind of research questions you can ask to analyze it. For example, you can ask the following descriptive questions, which can also help triangulate the results (add another layer of evidence):

• What are the most common words?
• What are the most common word pairs?
• Are there groups of words that are related together? What do those clusters look like?

On a more inferential sense, you can use text data as "quantitative" data. We will not go into the detail of this process here, but simply put, by looking at the occurrence statistics (e.g., TF-IDF stands for Term Frequency Inverse Document Frequency), we can see if they have predictive power over some outcome measures. For example, in a setting where you have a pre-post design to collect student knowledge gain data, but also captured students reflections along the way, you can ask the following question.

• What are the words that predict student knowledge gains?
• Silge's YouTube videos can give us some example questions too.

Using student exit ticket reflections for "teaching analytics" and fidelity (e.g., "what did you learn today?" -> analyze text data to see what the students perceived that day's lesson to be about).

Analyses

Now that you have a research question, let's see what we need to analyze our data.

• dictionary based
• frequency based
• lexical features (complexity scores)
• topic modeling
• Supervised Machine Learning

DESCRIBE THE R WORKFLOW UNTIL RESULTS ARE PRODUCED. Including graphs etc.

The Write Up

We know have a very interesting research question, good data to analyze, and our results of the data analyses at hand. How do we write our key sections for our paper:

• Methods
• Results
• Discussions is up to you :)

Sample Data Analyses section for Methods for Text Analyses

To analyze student reflection data, we first......Then, using XYZ package (CITE), we conducted xyz analyses.

Results

We conducted XYZ analyses to understand what features of student reflections predicted their knowledge gains. Our results indicated that XYZ. These are the important statistics to report and this is what they mean.