HOMEWORK #2 Ideal data

(see Syllabus for due date and points possible)

Objective

Locate a news article that describes a test of an abstract, causal model (e.g. smoking causes cancer; football festivities on a weekend cause a drop in student attendance the following Monday; the topic need not be biology or one of the "natural" sciences). From the description contained in this article, identify the goal of the study, one abstract model being tested (e.g. "caffeine is addictive" or "exposure to some pesticide causes increased risk of breast cancer"), and fill in our ideal data template for the model you choose, as given below. Most importantly, the article needs to refer to some data used to test the model. Most of this assignment is about the data.

A detailed example of the "ideal" homework assignment follows. This format should be followed as closely as possible. Note that for each feature of the ideal data template, the type of answer that is expected is indicated. Especially note the following:

i)               Your grade on this assignment will depend on the suitability (e.g., depth) of the article you chose. In general, the data should meet at least three of the properties of ideal data for full credit.

ii)             the article must be less than 9 months old and should not be from an on-line quiz.

iii)            Upload your filled Word form (it's OK to convert it to pdf) to the Bblearn account for Homework 2; if your Word file does not provide a link to your article, upload the article separately (a copy is OK)

 

How to use the Data Word template (for Homework 2)

Name: Type your name in the field.

Article: Type the title, source, and date of the article in the appropriate fields . Either provide a link to the article or upload the article (as a scan, if need be). We want to be sure we have access to the article, so a little redundancy does not hurt.

I. Goal:

In one sentence or less, describe the goal of the study in your article.

II. Model Being Tested

Must be a causal, abstract model -- such as a hypothesis – a model for why or how something happens (e.g., smoking causes lung cancer).  If the model is not causal, you should find one that is, or change the article.

Be sure to get this part right.

In one sentence or less, state the abstract, causal model (e.g. the hypothesis being tested). Although the article may make reference to many models, the model you choose here must be the one that the data are testing -- a common mistake has been to choose a model that is referenced in the article, but is not the reason the data were gathered. Another common mistake has been to use an article that never offers data but just talks about the conclusions of a study. It may be easy to identify the model in this case, but the assignment will get a low score if there are no data mentioned. Indeed, and as also mentioned below, the nature of the data needs to be described in enough detail that you can identify at least 3 ideal features as present.

III. Design Feature:

Using the format specified below, fill in the following information for each of the five design features of ideal data.  You will want at least 3 of the design properties to be “present” for full credit:

1. Relevant: Considering the goal and model you have identified, is this design feature relevant to the study? That is, should the study have included this feature, regardless of whether the study did include it? If so, answer the next two points as well.

 

2. Status: In one word state whether this particular design feature is present, absent or ambiguous/uncertain. Ambiguous (uncertain) means either that the article did not state explicitly that this particular design feature is present at any level or absent, or that it is not possible to infer from what is stated whether the feature is present or absent.  If the feature is present at any level, it counts as present.

 

3. Quotes: Quote the article directly. Copy from the article the sentence or phrase that either describes the specific design feature, states that this design feature is not present, or indicates ambiguity. Do not include more than is necessary.

Follow this format for each of the 5 ideal data features:

A. Explicit Protocol

1. Relevant:

2. Status:

3. Quotes:

B. Replication

1. Relevant:

2. Status:

3. Quotes:

C. Standards (used to assess data quality)

1. Relevant:

2. Status:

3. Quotes:

D. Randomization

1. Relevant:

2. Status:

3. Quotes:

E. Blind

1. Relevant:

2. Status:  (treat a “placebo” as evidence of blind)

3. Quotes:

 

Copyright 1996, 1997 Craig M. Pease and James J. Bull. All rights reserved.
cors236@uidaho.edu