Statistics of events and invitations

Question

I am going to be hosting a number (~10) of potluck meals over the course of the summer, my pool of people to invite is about 40 people with about 10-15 coming to each meal. So I figure this would be a good opportunity to record data over time about the meals/people. The issue I am having is I am not sure what information to keep track of and what format to record it in.

Here are some examples of trends I think would be interesting:

1. How many meals I have invited people to
2. On average which round of invites did people get invited to (some people rsvp as no in the beginning and so there is another 'round' of invites)
3. How many meals people have attended
4. What items people have brought

I have started a spreadsheet where each page is a meal, the first few columns of the page represent different rounds of invites, I input a persons name in the column that corresponds to the round of their invite. The last two columns are the ultimate rsvp from any round of invitation and the item brought if applicable.

To summarize I am looking for an efficient and concise way of recording the data associated with these meals for the trends mentioned. Additionally I am looking for other trends I can keep track of, I am doing a lot of this communication via email so timestamps would potentially be available for other interesting trends.

Help with good tags for this question would be appreciated.

Answer 1

Yoel, great question! I will address your question of what can be an "efficient and concise way of recording the data". Given your small data set, the following thoughts are more of theoretical nature than of practical use.

You have (what social scientists call) a multilevel data set, e.g. students (level 1) are nested in classes (level 2) which are nested in schools (level 3).

Unfortunately, your case is more complicated because each meal can be attended by more than one person and each person can attend more than one meal. So, there is no easy way to handle $1:n$ relationship between meals and persons. Furthermore, it is unrealistic to assume that each person brings unique items to each meal, i.e. many will bring salad, bread, cheese, etc. (that's at least my experience ;-) Again, there is no easy $1:n$ relationship between person and item(s). The following picture may be helpful to illustrate what I mean:

If you had more data, I would recommand using a relational database management system (MS-Access, MySQL, SQLite etc.) and you would need the following 5 tables (or relations). Each table needs at a minimum the following identifier variables:

Meal:

• meal.id

Person:

• person.id

Item:

• item.id

Since you do not have $1:n$-relations, you also need auxiliary-tables which help you to establish $n:m$ relations:

meal-person:

• meal.id
• person.id

person-item:

• person.id
• item.id

By the way, if you intend to do more complicated (regression) analyzes than those mentioned in your question, then you will need to run "Multilevel Models With Crossed Random Effects".

Update

Yoel asked for some resources to get started with multilevel analysis:

• My all time favourite intro text is "Using SAS Proc Mixed to Fit Multilevel Models, Hierarchical Models, and Individual Growth Models" by Judith Singer. This is a basic introduction to multilevel models. The text seems, however, currently not available.
• The University of Bristol Centre for Multilevel Modelling offers "Training materials and online information about multilevel modelling and MLwiN". Lean back and enjoy the videos!
• Joop Hox has written one of the best textbooks and a lot of articles. Some of them can be found on his homepage.
• The UCLA Academic Technology Service offers textbook examples, see the section about Multilevel Modeling.
• Douglas Bates is currently working on a book called "lme4: Mixed-effects Modeling with R". Chapter two is about "Models With Multiple Random-effects Terms".

However, I was not successful in finding a good intro text which covers multilevel models with crossed random effects. Maybe, this blog post can get you started.