Representing the Twittersphere: Archiving a representative sample of Twitter data under resource constraints

The rising popularity of social media posts, most notably Twitter posts, as a data source for social science research poses significant problems with regard to access to representative, high-quality data for analysis. Cheap, publicly available data such as that obtained from Twitter's public application programming interfaces is often of low quality, while high-quality data is expensive both financially and computationally. Moreover, data is often available only in real-time, making post-hoc analysis difficult or impossible. We propose and test a methodology for inexpensively creating an archive of Twitter data through population sampling, yielding a database that is highly representative of the targeted user population (in this test case, the entire population of Japanese-language Twitter users). Comparing the tweet volume, keywords, and topics found in our sample data set with the ground truth of Twitter's full data feed confirmed a very high degree of representativeness in the sample. We conclude that this approach yields a data set that is suitable for a wide range of post-hoc analyses, while remaining cost effective and accessible to a wide range of researchers.     

基于排序集样本的Bayes统计推断

从参数估计的角度,讨论了基于排序集抽样的Bayes估计,并进一步与传统的Bayes估计相比较,从估计的相对效率角度说明了基于排序集抽样的Bayes估计的优良性。     

Bias and the limits of pooling for large collections

Modern retrieval test collections are built through a process called pooling in which only a sample of the entire document set is judged for each topic. The idea behind pooling is to find enough relevant documents such that when unjudged documents are assumed to be nonrelevant the resulting judgment set is sufficiently complete and unbiased. Yet a constant-size pool represents an increasingly small percentage of the document set as document sets grow larger, and at some point the assumption of approximately complete judgments must become invalid. This paper shows that the judgment sets produced by traditional pooling when the pools are too small relative to the total document set size can be biased in that they favor relevant documents that contain topic title words. This phenomenon is wholly dependent on the collection size and does not depend on the number of relevant documents for a given topic. We show that the AQUAINT test collection constructed in the recent TREC 2005 workshop exhibits this biased relevance set; it is likely that the test collections based on the much larger GOV2 document set also exhibit the bias. The paper concludes with suggested modifications to traditional pooling and evaluation methodology that may allow very large reusable test collections to be built.

Biased sampling activity: an investigation to promote discussion

The statistical concept of sampling is often given little direct attention, typically reduced to the mantra “take a random sample”. This low resource and adaptable activity demonstrates sampling and explores issues that arise due to biased sampling.     

Pigeons,facebook and the birthday problem

The unexpectedness of the birthday problem has long been used by teachers of statistics in discussing basic probability calculation. An activity is described that engages students in understanding probability and sampling using the popular Facebook social networking site.     

Television and the quality of family life

One hundred seven adults' family activities and experiences were studied via the Experience Sampling Method (ESM). Respondents carried automatic paging devices and were randomly signalled to fill out self‐reports over the period of one week. While some scholars have claimed that communication during television viewing is nonexistent, talking occurred during 20 percent of family television time. Respondents who watched more television also spent more time with their families and reported feeling as well affectively while with family members as did light viewers. However, family viewing experiences were reported as being more passive than family activities generally. Heavier viewers also reported lower levels of activation during non‐TV family activities than did light viewers. Implications are discussed.     

我国科技工作者参与科普活动的现状与障碍——基于第二次全国科技工作者状况调查数据

本文利用“第二次全国科技工作者状况调查”数据,分析了我国科技工作者参与各类科普活动的现状,发现半数以上科技工作者2007年参加过某种形式的科普活动,但参与水平仍明显低于发达国家同行,其中35岁以下青年科技工作者以及低学历、低职称科技工作者的参与水平更低.分析结果还显示,参与渠道、时间精力、相关经费、科普能力和激励动力的缺乏是制约科技工作者参与科普活动的主要障碍.在此基础上,文章提出了促进科技工作者进一步参与科普活动的建议.     

山东省30岁以下独生子女比例调查的抽样设计

随着我国独生子女群体年龄的增长和规模的扩大,该群体引发的社会学和人口学问题越来越成为人们关注的热点。独生子女在同龄人中的比例大小对一个地区整体家庭户的规模和人口结构的变动以及总和生育率有很大影响。因此对该群体的调查十分必要,文章主要依据山东省的人口分布和行政区划特点对山东省30岁以下独生子女比例调查进行科学合理的抽样方法和抽样方案的选择和设计,并给出了回推总体的方法。     

Predictive validity evidence for an assessment program based on the Work Sampling System in mathematics and language and literacy

This study examined the predictive nature of teacher ratings of student performance on the mathematical thinking and language and literacy domains of a state mandated curriculum-embedded performance assessment for children in first grade to student achievement on a criterion-referenced test in third grade in mathematics and English language arts, using a multilevel modeling approach. The sample included 1,281 elementary school children nested within 132 classrooms in an urban school district in South Carolina. Results showed positive associations between first grade teacher ratings and students’ third grade achievement. However, first grade teacher ratings were not found to uniquely contribute to the proportion of explained variance in third grade performance on a high-stakes statewide assessment, over and above student demographic variables.