Review (5-10 min)
Since we haven’t been together as a group in about a week, I want to briefly review the first two weeks.
- Rhetoric and Data: With rhetoric, we are interested in the intentional use of symbols, and in relation to data, that means a few things. First, we use symbols to make data–how they are categorized, represented within those categories, manipulated by calculations to transform these data. Then we talk, write, and draw things that represent a summary of characteristics of how we create and transform data. (see PowerPoint on CourseWeb that we talked about on 8/27).
- Finding data: what do you want to know? how can you get data about it? what organizations or data repositories might have something here? Will you need to do your own collection? What will you have to do to collect useful data? (see lesson plan from 8/29 and 9/3 lesson plan)
- Contextualizing data: How was it collected? How were categories defined? What can you say and not say about this kind of data (e.g., media reports will create a lot of duplicitous data if you are interested in separate events)? What are its limitations? What isn’t reflected in the data that you might have to acknowledge (e.g., people not part of the dominant culture may not voluntarily engage with institutions doing data collection)? (see chapter 2 of Data Feminism and 8/29 lesson plan)
- Cleaning data: Is everything standardized to avoid inaccurate counts? Is everything accurately recorded? Are their duplicates? Is there anything wrong with any columns or rows that would interfere with techniques of analysis? Is there missing data and should you do something about that? (see 8/29 lesson plan for more)
- Exploring data: Looking through the different columns and rows you have, what jumps out at you? Can you do some simple arithmetic to see what is more or less than you expected? Can you create a histogram or a box plot for numeric data to show the distribution of values? Is the data very varied or clustered together around the same point? What is the mean, median, or mode? For categorical data, what is the distribution of values? A lot of one type? Can you find two variables and create a scatterplot to see if there is a relationship in any direction? A bar chart for categorical data or a pie chart to see proportions of a variable? Finally, when finding something to write about, something “we already know” and something that might at first seem “not useful or interesting” might be initial impressions that are not accurate. Redundant information might actually be friendly with creative but repetitive writing. Something not interesting at first, might have layers to it not seen initially. All this to say, don’t move too fast! (see 8/27 lesson plan on data types, 8/29 and 9/3 for steps to explore data)
- Writing about data (so far): There are many aspects of writing with data that we are going to explore this term. On 9/5, we got a chance to think about the general summary of what we will look at: the nature of prose vs. images, design and accessibility, the organization of the argument or narrative, interaction between prose and images, how language and images can simplify complex information (e.g., examples, metaphors), how methods and limitations are discussed for how data were generated and analyzed, how evaluation of the number is communicated, and how writing constraints interact with any of the previous (e.g., genre of the writing, audience, print vs. digital media). See “questions to ask as you get started writing” in the 9/5 lesson plan, adapted from chapter 1 in Miller. (see 9/5 lesson plan).
- Public Writing: For public writing, you have to consider expectations that lay audiences might have for more technical or scientific information that data analysis tends to help communicate. Miller pointed to this general organizational trend: be up front about what question you are exploring, the data you look at, how you explored it, and what application or action comes from this. There needs to be a story, something interesting about it. Genres include: issue briefs, white papers, brochures, news articles, blog posts, websites, factsheets, chartbooks, public reports. (see 9/5 lesson plan)
Miller’s 7 Principles (30-45 min)
To quickly re-familiarize ourselves with chapter 2, let’s paraphrase each principle. I’ll break you into seven groups. In your pairs (and in two groups of three), put each principle in your own words corresponding to which principle you receive. Paste your principle into this Google Doc.
Example Texts
Okay, now that we have these re-familiarized, let’s look at a few examples to see how much or how little they adhere to these principles (NOTE: not each principle will apply to every statistic). In your group/pair, apply each principle to see if it is present.
First up: a recent blog post on YouTube’s official site about “removing harmful content.”
Here is the statistic to analyze, but see full blog for checking against all principles. Also consider if the evaluation presented or implied here is fair:
For example, the nearly 30,000 videos we removed for hate speech over the last month generated just 3% of the views that knitting videos did over the same time period.
Now, I want you to compare two passages using the “same” statistic. How do these both hold up against the principles? Do they come to the same evaluation? Are they different? Are they both fair?
The first comes from Yahoo! Finance about the December 2017 jobs report.
The labor market ended 2017 with a slight disappointment.
The December jobs report from the Bureau of Labor Statistics showed nonfarm payrolls grew by 148,000 while the unemployment rate stayed steady at 4.1%, slightly missing expectations.
Economists expected nonfarm payrolls grew by 190,000 in December while the unemployment rate was expected to remain at 4.1%. The unemployment rate is currently at the lowest level since December 2000.
The second comes from CNN Money about the same report.
The U.S. economy added 2 million jobs in 2017, another solid year of gains.
In December, the economy added 148,000 jobs, according to Labor Department figures released Friday. That was below what economists expected, but still the 87th straight month of gains — the longest streak on record.“The 2017 job market was really great,” said Cathy Barrera, chief economist at ZipRecruiter, a jobs website.
Unemployment remained at 4.1%, matching the lowest level in 17 years.
Try it out
Now that we have the sense about the principles, and especially about how evaluation of a number (that is, how writers signal interpretations), let’s try something out in the pair or group of three you are currently in. Here is the statistic you’ll work with:
9% of plastic has been recycled.
This statistic comes from a paper in Science Advances. Going through the 7 principles from Miller (or, the ones that apply), write the “lead” paragraph of a news article (pages 2-3 from the Science Advances article might be helpful here to help add to this). Think about how you will signal an evaluation of this number (this can be rather subtle or very explicit).
A lead paragraph is just the opening paragraph in a news article that gives the main ideas of the piece (you can look at the links to the example articles on the job report above or the data journalism examples from last week for a model). If you need more room than a lead paragraph would provide, feel free to “continue” the article on in another paragraph or two.
Once you are finished, I’m going to pair you with a different group so you can compare your versions. If time, try to use both versions to create one “revised” version based on the elements you like best from each.
Once finished (either the “Revised” version, or if we don’t have time, the unrevised version), let’s post them up on the same Google Doc where we have Miller’s principles listed.
Other considerations
- What you are talking about should influence how you communicate your interpretation. If you have a statistic about human suffering vs. a statistic about, say, plastic, should you take a different approach? How so? How does this influence the “evaluative” moves you make as a writer?
- There’s a balance between “checking boxes” in the way that Miller’s principles are structured and doing what is most important about writing with data: foregrounding what is interesting and important about the analysis.
Design and Accessibility (5-10 min)
On CourseWeb, there is a handout for things to consider about design and accessibility that can help you as you start working on your first drafts. We will talk about all of these elements throughout the term, but I wanted you to have a general sense early on as you start to write.
What’s Next (5 min)
-first draft of public writing project is due 9/19.
-next class, we will talk about more technical concerns about writing with data. You’ll read chapter 4 of Miller, for more principles.