-
Notifications
You must be signed in to change notification settings - Fork 0
/
.html
81 lines (69 loc) · 4.53 KB
/
.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
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
<!DOCTYPE html>
<html lang="en">
<head>
<title>Compumetrika</title>
<link href='http://fonts.googleapis.com/css?family=Open+Sans:400italic,600italic,700italic,400,600,700' rel='stylesheet' type='text/css' />
<link href='http://fonts.googleapis.com/css?family=Merriweather:300' rel='stylesheet' type='text/css'/>
<link href='http://fonts.googleapis.com/css?family=Source+Code+Pro:200,400,700' rel='stylesheet' type='text/css'/>
<link rel="stylesheet" type="text/css" href="/theme/css/icons.css"/>
<link rel="stylesheet" type="text/css" href="/theme/css/styles.css"/>
<meta charset="utf-8" />
</head>
<body id="index">
<!-- header -->
<header class="siteheader">
<!-- site image -->
<div class = "sitebanner">
<h1><a class="sitetitle nodec" href="">Compumetrika</a></h1>
<h3 class ="sitesubtitle"></h3>
<!-- nav -->
<nav class="menu">
<ul>
<!-- menu items-->
<!--pages-->
<li><a class="nodec" href="/pages/about.html">About</a></li>
<li><a class="nodec" href="/pages/advice.html">Advice</a></li>
<li><a class="nodec" href="/pages/contact.html">Contact</a></li>
<li><a class="nodec" href="/pages/reseach.html">Research</a></li>
<li><a class="nodec" href="/pages/resources.html">Resources</a></li>
<li><a class="nodec" href="/pages/teaching.html">Teaching</a></li>
<!-- services icons -->
</ul>
</nav>
</div> <!-- sitebanner -->
</header>
<!-- content -->
<section class="content">
<h3 class="posttitle">
<a class="nodec" href="/.html" rel="bookmark" title="Permalink to Current Reading">
Current Reading
</a>
</h3>
<div class="postinfo">
<p class="published" title="2014-09-15T00:00:00">
Mon 15 September 2014
</p>
</div><!-- .postinfo -->
<div class="article">
<p>Finished an experimental paper I've had on the back burner: [...]</p>
<p>It touches on a perinial question for anyone using surveys: how do you handle free-form responses? </p>
<p>Free-form response gives experimental subjects the most freedom in expressing thought -- one can perhaps avoid "contaminating" subject responses with the exerimenter's own desired outcomes. </p>
<p>(In a more macro-oriented context, this was a motivating theme of Blinder's "Asking About Prices," as well as Bewley's more recent (and still unpublished?) "interviews with pricing managers" project. The authors found some disconnect between the categories of "reasons for a price change" dictated by theory, and the reasons provided by industry professionals when they were allowed free-form response.)</p>
<p>Of course the trouble with free-form response is that it is not very amenable to quanititative analysis. Formally categorizing human language is notoriously difficult -- Chomsky's "failed" attempts still produced quite a lot of theory foundational to modern computer science ("This is the <em>science</em> of computer science!" as my theory professor used to say). Even aside from theoretical difficulties, if the experimenters/researchers sift and categorize all the responses themselves, this immediately invites the specter of "selecting your results." </p>
<p>In practice, "content analysis" is the main option - some trained analysts sit in (perhaps different) rooms and manually code responses. </p>
<p>This experiemental paper presents an alternative: instead of typical content analysis, the authors use game theory to set up a coordination game on content of the free-form messages. In a sense, this is a formalization of the idea of "intercoder reliability" from tranditional content analysis. </p>
<p>One wonders if this concept could be "taken one step higher," in the sense that categories themselves are choosen by coodination. Perhaps the first round of coordinating subjects chooses a set of N categories, and a second round of subjects assigns messages to the categories. This is almost certainly more difficult, but perhaps an apporpriate framing could allow it. (Perhaps 10 mote minutes of thinking about it will convince me this isn't useful...)</p>
<p><a href="./..">Home</a></p>
</div><!-- .content -->
</section>
<!-- footer -->
<footer>
<p>
© Nathan Palmer, license <a href=""> </a>
unless otherwise noted.
Generated by <a href= "http://docs.getpelican.com/">Pelican</a> with
<a href="http://github.com/porterjamesj/crowsfoot">crowsfoot</a> theme.
</p>
</footer>
</body>
</html>