Module 5: Library data literacy workshop

View study sheet (PDF) View SQL cheat sheet (PDF)

Teaching this module? A facilitator guide adds a preparation checklist, timing cues, discussion guidance, slides, and a printable PDF.

This workshop is a general-audience introduction to reading and questioning a real dataset, designed for a public or academic library setting rather than a credit-bearing course. Participants need no background in finance, statistics, or computers beyond comfort using a web browser. Working entirely in a no-code sandbox, they explore a dataset of auction and resale transactions for watches, jewelry, handbags, and other luxury goods, drawn from more than 850 vendors with history to 1949, and practice the basic moves of data literacy: asking a question, finding the right table and fields, running a filter, and reading the result honestly, including its limits.

Target course(s) and level

Public library workshop series, academic library continuing-education session, or community data literacy program. Appropriate for adult patrons and lifelong learners with no assumed background; also usable as a library instruction session for undergraduates outside a formal course.

Learning objectives

By the end of this session, participants will be able to:

  1. Navigate a real, professionally maintained dataset using a point-and-click browser tool, with no coding required.
  2. Explain what a data dictionary is and use one to understand what a field in a dataset actually measures.
  3. Filter a dataset to answer a simple, self-generated question (for example, what has a particular brand sold for recently).
  4. Read a chart built from filtered data and describe, in plain language, what it does and does not show.
  5. Identify at least one reason to be cautious about a data-driven conclusion, such as an incomplete category or an unstable recent period.

Prerequisites

None. Comfort using a web browser and typing into a search box is sufficient. No prior exposure to spreadsheets, databases, or SQL is assumed; the sandbox's pre-built charts and filters are used in place of any query language.

Materials and access needed

  • Sandbox access at sandbox.altfndata.com. Each participant self-registers with an email address (a personal email is acceptable for this general-audience session, though a work or school email speeds automatic approval); library staff should confirm in advance whether the venue's guest wifi and registration flow work smoothly together.
  • A projector or large screen for the facilitator demo, plus library computers or personal laptops for participants to follow along.
  • The coverage browser and data dictionary tabs, which are the primary tools used in this workshop rather than the SQL editor.
  • No instructor class API key or coding tools are needed for this module.

Session outline (60 minutes)

  • 0 to 10 min: Welcome and framing. What is a dataset, and why would anyone want to look at auction records for watches or handbags.
  • 10 to 20 min: Sandbox orientation. Everyone registers and opens the sandbox together; facilitator walks through the homepage, coverage browser, and data dictionary.
  • 20 to 40 min: Guided demo. Facilitator walks the group through filtering one brand's sold items and building a chart.
  • 40 to 50 min: Participant exploration. Each person picks a brand or category they are curious about and filters to it themselves, with facilitator circulating to help.
  • 50 to 60 min: Group share-out and discussion questions.

In-class demo (sandbox-first, no code)

  1. Open sandbox.altfndata.com and register with an email address. Approval happens automatically, so everyone should be in within a minute or two.
  2. Click the coverage browser tab. Type a familiar brand name, such as a well-known watch or handbag maker, into the search box and note how many records and vendors appear.
  3. Click the data dictionary tab and find the entries for sale_date, usd_price_decimal, and status. Read the plain-language description of each aloud, so participants understand exactly what they are about to see.
  4. Return to the coverage browser or filter panel and narrow the dataset to that same brand, restricting to sold items only.
  5. Look at the resulting list of items: note the item title, the sale date, and the realized price in US dollars for a handful of entries.
  6. Open the pre-built charts tab and select a chart that shows realized price over time for the filtered brand.
  7. Ask the group: what does this chart tell you, and what does it not tell you? Point out that the chart shows what buyers paid at auction, not the brand's retail price or overall popularity.
  8. Use the CSV export button to show participants how they could save this filtered view for their own reference, and note that the same filtered view is available any time they log back into the sandbox.

Datasets and queries used

Dataset: handbags data or watches data, accessed through the sandbox's coverage browser and filter panel rather than the SQL editor. For facilitators who prefer to demonstrate the underlying SQL editor as a bonus (optional, not required for this audience), the equivalent filter is:

SELECT designer, item_title, sale_date, usd_price_decimal
FROM all_handbags_data
WHERE designer LIKE '%Hermès%'
  AND status = 'sold'
ORDER BY sale_date DESC
LIMIT 25;

Fields referenced during the workshop (all shown through the coverage browser and data dictionary, in plain language): designer, item_title, sale_date, usd_price_decimal, status, vendor.

Discussion questions

  1. What surprised you about what this dataset actually contains, compared to what you expected before we opened it?
  2. Why does it matter whether a listed price was what an item actually sold for, versus what a seller merely asked for it?
  3. What questions would you want to ask about how this data was collected before trusting a conclusion drawn from it?
  4. If you saw a chart showing prices rising for a brand over the last few months, what would make you trust that trend, and what would make you doubt it?
  5. How is looking at real auction records different from reading a news article that summarizes a market trend?
  6. What is one everyday decision, unrelated to finance, where being able to filter and read a real dataset yourself would be useful?

Homework assignment

This module does not carry a graded homework assignment, since it is designed for a general-audience library setting. Facilitators who want an optional take-home activity can ask participants to return to the sandbox on their own, pick a category they did not explore in the session (for example, fine art or wine and whisky if the session covered handbags), and bring one observation and one question about the data to a follow-up session or a brief written note left with library staff. There is no formal deliverable, and the activity is intended to reinforce independent, unsupervised comfort with the sandbox rather than to produce a graded product.

Going deeper

Key terms

  • Field: a single column of information in the dataset, such as a sale date or a price, defined plainly in the data dictionary.
  • Filter: a rule that narrows the dataset to only the rows meeting a condition, such as one brand or one status.
  • Data dictionary: a plain-language guide to what each field actually records, used before trusting any number pulled from it.
  • Coverage browser: the sandbox tool for checking how many records and vendors exist for a brand or category before filtering to it.
  • Status: the field marking whether a listed item actually sold or went unsold, the difference between an asking price and a real one.
  • Realized price: what a buyer actually paid for an item, as opposed to what a seller asked or an auction house expected.
  • Vendor: the auction house or platform where an item was sold.
  • Recency gap: the tendency for the newest few months in the dataset to look thin simply because collection has not caught up yet, not because the market slowed.

Common pitfalls

  1. Treating an asking price as if it were a sale price, rather than checking status for "sold" first.
  2. Reading a dip in the most recent months as a market trend, when it is usually a recency gap in data collection.
  3. Trying a brand name once in the coverage browser, finding few results, and concluding there is no data, instead of trying a shorter or differently spelled version of the name.
  4. Reading a chart of one filtered brand or category as if it described the whole market, rather than just that slice.
  5. Forgetting what filter produced an exported CSV, so the saved file becomes hard to interpret later.
  6. Assuming a higher record count for one brand means it is more popular, rather than simply better covered by the auction houses in the dataset.

Additional queries to explore

  1. Coverage browser task: look up two familiar brands in the same category (for example, two watch brands) and compare their record counts and number of vendors, to see which one the dataset covers more thoroughly and why that matters before drawing conclusions about either.
  2. Pre-built chart task: build the sold-item-count-by-quarter chart for a category you did not use in the demo, find the most recent quarter, and discuss as a group whether that quarter should be trusted given the recency gap.
  3. Data dictionary and filter task: read the plain-language definition of vendor, then filter to sold items from one specific auction house and note how the results differ from filtering by brand alone.

Extension activities

  1. Ask each participant to explore a category not covered in the session, such as fine art or wine and whisky, and bring back one thing that surprised them.
  2. Have participants write a single sentence describing a limitation of the dataset they would want a reporter or a friend to know about before repeating a number from it.
  3. For a curious participant or two, show the optional SQL editor version of the filter used in the demo, purely as a preview of what Module 2 covers in depth.

Connections to other modules

Module 2, data science and SQL, teaches the query-language version of the same filtering and reading skills practiced here by hand in the coverage browser. Module 3, consumer and luxury economics, picks up the brand-level pricing questions this workshop only opens the door to. Module 13, data ethics, quality, and coverage bias, goes much deeper into the recency gap and other reasons to read a dataset cautiously.