Some miscellaneous comments that arose in several of the projects:

(1) getting the raw data from online survey services

    if you are using surveymonkey.com, questionpro.com,
    surveygizmo.com, etc., please make sure you can download the raw
    response data in an xls spreadsheet.  Merely using the free
    summaries will not be sufficient for the class, because the free
    summaries do not give you enough flexibility to explore
    *relationships* between the variables (e.g. do freshman women find
    the UC gym facilities more attractive than junior men?).

(2) handling nonresponse

    (a) even if you are doing a multi-phase or multi-pool survey, it
        is better to do some reminding in the earlier pools before
        moving on to the later pools.  Otherwise all you are getting
        is the 'easy responders' in each pool: You will never get
        opinions from people who look more like non-responders, and
        therefore your results will be more susceptible to
        (self-)selection bias.

    (b) many web survey providers will either (i) give you a list of
        who has not responded to the survey invitation yet; or (ii)
        send reminders to nonresponders directly without giving you a
        list of who they are.  This is usually better than sending out
        blanket reminders to everyone invited, since then you won't be
        irritating people who have already responded to your survey.

    (c) Some groups are considering Google Docs Forms instead of a
        regular web survey provider.  There are two problems with
        this, as far as I can tell:

        (i) You don't have any way to prevent the same person from
            filling out the form more than once (or 100 times),
            biasing your results.

        (ii) You don't have any way of knowing who has not filled out
            your survey yet, so reminders would have to go to
            everyone, not just non-responders.

        There may be ways around these problems (I don't know!), but
        in any event, please do *not* use Google Docs unless you've
        come up with solutions and cleared them with me.

(3) consent forms

    If respondents have questions about their rights as research
    participants or to report objections to the study, please have
    them contact me rather than IRB, since there is no record of your
    study with the IRB.

    Also, remove statement that IRB has approved the study.  Instead
    add the statement: 

    "This research project is exempt from Carnegie Mellon University
    Institutional Review Board (IRB) review, since it is being
    undertaken as a course requirement."

all best,

-BJ