EM 2019 - Discussion Qs
 Share
The version of the browser you are using is no longer supported. Please upgrade to a supported browser.Dismiss

 
$
%
123
 
 
 
 
 
 
 
 
 
ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
Experimental Methods Fall 2019 - Discussion Questions
2
Week 2: Fundamentals of Causality and the Experimental Method
3
4
Instructions: Type your discussion question below beside your name. Discussion questions are due Monday at 3pm (the day before the in-class discussion). On Monday night or early Tuesday morning read through all of the discussion questions and upvote two questions you are particularly intrested in discussing. Add your vote by changing the number in the upvotes column + 1 (e.g. if the number is 0, change it to 1; if the number is 3, change it to 4; etc.).
5
6
7
StudentDiscussion QuestionUpvotes
8
Attara, HuisseinHolland (1986:7) mentions the importance of temporal stability and causal translence. How can we design a study investigating X in two countries (units) at different historical times inorder to make a comparison between the two units that might be undergoing similar scenarios? 1
9
Aweko, ScoviaHolland (1986:7) and Imbens & Rubin (2015:8) assert the importance of temporal stability. However, it is unclear how temporal stability is maintained in experiments where treatment is to be adminstered over a period of time to the same unit (especially human subjects) since we can not expect the unit to stay the same over time (if one is more sensitive to the treatment at one given time than the other). 2). To what extent does spillover influence the credibility of a natural experiment if the researcher can't effectively control/ minimize for it as Imbens & Rubin (2015:9) suggest?1
10
Beltran, FelixHolland (1986:11) mention natural experiments by stating that "For example, in the medical and social world we might be able to conceive of an experiment, but no one would ever try to carry it out. Instead, we might have to wait for a 'natural experiment' to occur. Observational
study is the term used by statisticians (e.g.,Cochran 1983) to refer to studies for which 'The objective is to study the causal effects of certain agents' but For one reason or another the investigator can not . . . impose on ... or withhold from the subject, a treatment whose effects he desires to discover." My question is, how can we replicate, or verify the external validity, of a natural experiment, given that most, such as the outcomes of natural disasters in a study, are almost never identical, such as the different variables within a natural experiment study?
3
11
Frase, RobHolland “take[s] the position that causes are only those things that could, in principle, be treatments in experiments” (1986:954). Do you accept this stance? Do you accept Greiner and Rubin’s rebuttal that we can manipulate perceptions of immutable characteristics? Why or why not? How can experimenters tell if their respondents' perceptions were manipulated correctly?2
12
Hu, JiayinImbens & Robin (2015:14) points out the missing data problem, that is the potential outcome associated with any alternative treatment is missing when a treatment is assigned to an individual unit. Similarly, the fundamental problem of causal inference identified by Holland (1986:947) raises the same question. According to these two readings, how can we address/overcome the problem? What implications can we learn from the solutions? 5
13
Kim, HeewonGreiner & Rubin (2011) makes the rationale for using perceived immutable characteristics by asserting the importance of the ability to manipulate a treatment for experimental studies. However, I believe that there are some certain mutable things in our society that are hard to be manipulated but still are easy to be measured, other than immutable characteristics. For example, percevied social class, which was not included in our readings but I believe this is someting mutable, could not easily be manipulated, but still could be measured. Would comparing such mutable characteristics of subjects could also be considered as a cause for experimental studies?3
14
Kincaid, ReilyThe stable unit treatment assumption, ‘SUTVA,’ requires that units do not interfere with another and that for each unit, there is a single version of each each treatment level (Imbeds & Rubin, 2015, p. 10). Why would “redefining the represented treatment levels to comprise a larger set of treatments” or “coarsening the outcome” help satisfy the SUTVA assumption? What might these two strategies look like in a different example than Imbeds and Rubin’s (i.e., not aspirin and headaches)?1
15
Kuffuor, OwuraGreiner and Rubin (2011) state the importance of conceptualizing treatment as occuring at the moment of first perception. This is because it allows for the capturing of variables whose values were determined by after that moment may be affected by the perception. Intutively i think of the timing of the treatment as the moment when treatment is administered. By this definition, why is it important to take note of the timing of the treatment and what injuries are likely to occur if the timing is not considered.?0
16
McPhail, BrianHolland argues that “the analysis of causation should begin with studying the effects of causes rather than the traditional approach of trying to define what the cause of a given effect is” (p. 959). What is the difference and why does it matter?3
17
Osorio Castro, ElizaGreiner and Rubin (2011) argue that a shift in focus from actual traits to perceptions of immutable characteristic would allow researchers to address some problems common within observational studies, e.g. post-treatment bias. A constructivist approach that disaggregates these variables in their constitutive elements might solve the problem with fixed or assigned conceptions. However, today social constructions tend to be more fluid than before and more amorphous, while socialization process could lead those perceptions in a specific direction, discernable by some of the elements involved, the continuous and faster social change turn exogenous effects on perceptions about these constructs an important element. How do we address these endogenous and exogenous effects within this type of variables given the fluidity of the perceptions that we have today?
0
18
Sistek, HannaHolland (1986: 947) mentions two solutions to the fundamental problem of causal inference, the scientific and the statistical. How do they differ and what caveats do they come with?0
19
Zaborenko, CallieImbens & Rubin (2015:11) state that the second component is that there are to be no treatment variations. Let’s say there was an experiment on loneliness that required multiple measurements across time. Some of those who received a treatment also had access to more social partners than others during the time between measurements. This may not be easily or accurately disclosed. Those who received both the treatment and had partners showed marked improvement, but it was due to the combination of those factors. Is SUTVA a realistic assumption for anything beyond experiments that are artificially set up and rigidly controlled?0
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
Loading...