Suggested Content for Raw Data and Prospective Analysis Section

  1. What will your data look like? First, who, what, how many? Then, how could it be characterized in terms of cases x variables at the most abstract/simple level (e.g., "I interviewed 16 adoptees about their hobbies, educational plans, and favorite colors.").
  2. State your research question in terms of concepts (see this slide show on "conceptualization" for a reminder).
    1. If it makes sense, distinguish “dependent” from “independent” concepts, causes from effects, structures from behaviors from attitudes? Even if you are not doing a classic research design, talk about the mechanism of causes and effects that your data is "about."
    2. If you are looking at narratives and processes talk about them conceptually
  3. From concepts into variables (i.e., a thing that can take on different values)
    1. What are the variables associated with your concepts?
    2. How will you operationalize your concepts? How will you get from ideas to your actual fieldwork/data collection to a data matrix?
  4. B. What values can they take on?
  5. C. Measurable with single “question,” “item,” or “observation” or need a scale/index?
  6. 3. What variables do you expect you will use to define your population and sample? What variations in your independent variables do you want to be sure to have included in your study?
  7. A. What is going to be the best way to make sure you get this kind of variation? In other words, what are the pros and cons of different sampling strategies for your project?
  8. 4. What univariate frequencies do you expect to find? Try expressing these in terms of percentiles, means and modes and medians, dispersion, skewness, etc.
  9. A. Draw the pictures. Explain what they mean. Why are they shaped as they are?
  10. 5. What bivariate relationships are potentially of interest given your research questions?
  11. A. Sketch tables and scatterplots.
  12. B. Describe the quadrants.
  13. C. Show us how these illustrate the comparisons you are interested in.
  14. 6. Sketch the overall correlation matrix and fill it in with your expectations (e.g., +, -, or 0) of how you think your different variables are related.
  15. 7. How has this exercise caused you to change how you are thinking about your project?

My project is…

What will your data look like? First, who, what, how many? Cases x variables.

My project examines notification norms among college student peers. My data collection consists of asking 30 college students to tell me the story of times when someone was "left out of the loop," found out news in an inappropriate manner, got upset about how a notification was handled, told a friend whom they could/should (not) tell some piece of news, etc. Each informant yields about about three pages of notes and they recount on average 2 examples giving me about 60 "cases" overall. Based on my pilot study, I expect this will amount to about 90 pages of text.

What will you do with your raw collected data to convert it into analyzable data? I will process this text by reading through and making initial notes about each case of notification - who, what, etc. Then I will develop a more detailed coding scheme and tally all the cases into a cases by variables matrix. For each case I'll also pull out great quotes that illustrate various ideas and concepts. I'll either record these in a text field at the right hand side of a spread sheet containing all my data OR I will keep them in a text file where I will tag them with a case identifier and key words related to the concepts. I'll put the tags up front in a way that let's me sort the paragraphs based on tags.
What does your sample look like? What population is it drawn from? What selection biases are probably there? For this project I am "sampling" from among college students at Mills - that's a diverse student population but they are all students at a private women's liberal arts college. Most probably in early mid-20s. It's not, thus, a sample of general population or young people or women or students. I don't have a sense of how to describe the selection bias in terms of my concepts and variables (that is, I can't say that most are this or that or that I am probably missing some important category). That said, I realize this project is exploratory and concept/idea generating rather than trying to fix an answer to a question about the world.
State your research question in terms of concepts (see this slide show on "conceptualization" for a reminder).

My overarching concepts are social relationship, information disclosure/sharing, and norms. I think of social relationship in general sense of social distance - from highly intimate to highly anonymous. There are other variables connected to this dimension - exclusivity, frequency of contact, age of relationship, content of relation, etc. The behavior I am studying is the passing along of information. This can vary from "I'm getting married" to "the professor changed that assignment." What's important to me is that it is passing along of information that is or could be relevant to the recipient. That means that I can think of this in terms of deliberate and intentional passing of information - deliberate tellings, if you will, information given rather than information given off (Goffman 19xx). The third conceptual cluster has to do with norms and rules. In the prompt to my informants I ask them what obligation they are fulfilling or what expectation they are feeling or what rule they are following. Why, I ask, does it make sense that you were put out by not being told X?

distinguish “dependent” from “independent” concepts, causes from effects, structures from behaviors from attitudes

I am interested in two things. On the one side, I want to examine the connection between behavior and behavior expectations and the way people describe social relations. In this sense, the relationship is the independent variable. But I also hypothesize that the connection runs both ways - notification behaviors can restyle relationships (e.g., you THOUGHT you were best friends, but now you wonder….).

For each of my cases - an example of a notification situation - I will have some data on
What are the variables associated with your concepts?
  1. the informant's understanding of the relationships involved (concept: relationship)
  2. what the information was and how it was relevant to the parties (concept: information)
  3. (maybe) information on "who else knew" (concept: the telling)
  4. a narrative of the telling or not telling (concept: the telling)
  5. the informant's interpretation (concept: notification, meaning)
  6. (maybe) the longer term repercussions of the interaction (concept: effect? relationship)
What do you expect in terms of range of values of your variables? My zeroth order guess about the actual variation I'll observe in my variables looks like this:
  1. I think most of the relationships will be friends and family with maybe some work-related relationships.
  2. Since this sample is students I expect most of the shared information to be about school, relationships, and such. Maybe won't see much in the way of politics, serious money issues, organizational intrigue and the like.
  3. "Who else knew" might vary in a couple of different ways. On the one hand, expect numbers. On the other, the richest data is when people express the "who else" in terms of circles or categories. If I were doing this as interviews, this would be one of the places I'd probe, maybe even asking folks to draw a diagram of these.
  4. For the narrative of the telling, I will try to capture the story-telling aspect in the data. This is important because one of my "side theories" is about how telling about tellings is a way that norms get expressed and how we socialize one another in terms of our expectations (e.g., a good friend would not notify someone like that).
  5. I don't have a good sense for how to describe my anticipation of the "values" interpretations might take on. One thought is that some will interpret things as errors or clumsiness, some will see slights as intentional, some
Describe any published scales or indices you are using. My characterization of notification and variables is based on Ryan (2006). I am influenced by Bogardis' (year?) social distance scale work, but I'm not using it directly. Otherwise I am not using any existing scales, indices, or measurement tools.
What univariate frequencies do you expect to find? Try expressing these in terms of percentiles, means and modes and medians, dispersion, skewness, etc.

This is a bit of a stretch. I sort of have two "independent variables" (variation in the cases/informants : the type of social relationship and the type of information) and two "dependent variables" (how the notification happened and how it was interpreted). Then what I want in this section is to illustrate what I expect the distributions will look like.

Here's what I expect the univariate distributions to look like. The types of information or news that people were or were not notified about comes in a variety of "flavors" - there is a central tendency around type labeled around 5.5. The relationship types cluster around 3 - let's imagine that "good college friend." The distribution of types of tellings is bimodal - one cluster at the bottom are those where someone was left out while everyone found out and the other cluster is tellings where too many people found out or people found out too early. The interpretations run from revelation of how people really feel (around 0.0 on the chart) to friends being clueless to recognition of fragmented social circles.

univariate.png

The bivariate distributions show how the variables are related. Here we see type of news and relationship are somewhat uncorrelated. Both relationship and type of news form two distinct clusters with type of telling.

bivariate.png
// //

Extremely brief sketch