如何写 research summary -- 请大家来谈谈方法或者贴sample

galluplinc

这是个老问题，也是个常见问题。方法，篇幅，内容，都因人而异，也因时而异。总得来说，三种风格：写汇报，谈感想，讲故事。

前两种风格比较普遍，相信大家都很有经验了。我在这儿贴一篇我自己写的 类似于讲故事风格的summary。欢迎在读生们贴一些前两种风格的summary，这样可以给研究经历/背景偏弱的申请者们一些关于怎么写research summary的general idea。




Hinkin, T.R. (1998). A brieftutorial on the development of measures for use in survey questionnaires.Organizational Research Methods, 1, 104-121.

Published at the first issueof ORM (1^st editor: Larry Williams from Virginia Commonwealth,founder of CARMA, remember?)
1.     WHY did Hinkinwrite this article?
He saw the problem in theresearch, which was described by Schoenfeldt (1984) that “the construction ofthe measuring devices is perhaps the most important segment of any study. Manywell-conceived research studies have never seen the light of day because of flawedmeasures.” However, to use appropriate measurement and clearly address theconcerns and issues in one’s study has never been an easy job, especially whenthe construct itself is complicated (or, controversial in the research field,like how to measure LMX, uni-dimensional or multi-dimensional? Individual orgroup level? or multi-level? argued years after years with one agreementthough, which is “indeed controversial and worth further discussions”…) as wellas when the ways researchers approach to it draw different interpretations. Yet,on the other hand, it is also a common sense that the more abstract the construct,the more difficult it is for researchers to measure it. So, before making yourarticle sound, try making your survey design grounded, and reasonable, and convincing,and, hopefully, beautiful, or, at least, acceptable … …
2.     With WHY said, SO WHAT?
Given that we organizationalscholars rely heavily on the use of survey questionnaires, before firing out asurvey to bother the industry guys, we’d better make sure that (1) we ourselvesunderstand the construct we are going to measure (i.e., concept, antecedents, process,structure, outcomes, implications, etc.); (2) we know what kind of elements weneed in our survey to not only accurately but also appropriately measure theexact things we intend to look at (personal opinion: no trade-off at this stage);(3) it is at least somewhat possible to get the right (not the expected!!!) databack through our design; and (4) we think about how far we can go abouttrade-offs on the instruments we put out there as our spokesman to shape, represent,and defend THE construct we want to investigate.
3.     What did Hinkintry to sell?
Main menu: “to provide a conceptual framework and astraightforward guide for the development of scales in accordance withestablished psychometric principles for use in survey research … will discussthe development of measures consisting of multiple scales as well as singlescale with multiple items”
Target group: readers with limited knowledge or methodologicalexpertise in scale development but somewhat familiar with certain statisticalconcepts and methods (guys, for most of us, don’t even bother thinking about hewas referring to us … we have a long way to go to fall into this group … that’swhy we come and sit struggling in 905 … BUT, look around, Sam and Lucas, showtime!!!)
Preface: APA (1995) states that “an appropriate operationaldefinition of the construct a measure purports to represent should include ademonstration of content validity, criterion validity, and internalconsistency”. Together, these provide evidence of construct validity (answer theQ: are we measuring what we intend to measure?). Nunnally (1978) proposed threemajor aspects of construct validation: (1) specifying the domain of theconstruct; (2) empirically determining the extent to which items measure thatdomain; and (3) examining the extent to which the measure produces results thatare predictable from theoretical hypotheses.
On-sale items:  Scaledevelopment process for NEW measures
Step I: Item generation. What we need in advance is the well-articulatedtheoretical ground that can be employed to CLEARLY and SOUNDLY identify thedomain for the new measure. Not necessary to look at the complete domain ofinterest, but make sure that we are asking NCAA football team playersabout how football in US looks like VS. asking huskers whether collegefootball games drive girls crazy.
Methods & process: deductive(pull out enough info from theory; pro: help to assure content validity ifproperly done; con: time-consuming, require some knowledge about thetheoretical definitions of the construct), inductive (used whenconceptual basis of a construct would not result in easily identifiabledimensions for which items can be generated; ask a sample of respondents todescribe their feelings/behaviors; CFA skills required here), itemdevelopment (simple and clear wording/terming), content validityassessment (get rid of whatever item proved to be conceptually inconsistent).Ideally, end up with 4-6 items per scale.
Step II Questionnaire Administration. Select a right sample that could produce enoughvariance in responses and avoid the effects of an idiosyncratic context.
Step III Initial item reduction. EFA is used with key assumption that all itemsbelonging to a common domain should have similar inter-correlations. APAmandates the report of reliability matrix.
Step IV Confirmatory factor analysis. Glad you make it here. Good trip huh? Let’s check thequality of the factor structure. Non-significant (the smaller, the better fit) chi-squareis desirable (small differences between model-implied variance and covariance &observed variance and covariance). Pray to God that “round-trip ticket” wouldnot jump out and say hello to you.
Step V Convergent/Discriminant Validity. Double quality check (Toyota really should have donethis better!!!). Look at construct validity to see the extent to which monkey correlatewith orangutan (convergent) and to which it does NOT correlate with squirrel (discriminant).
Step VI Replication. Use two different samples for scale development and psychometricproperty assessment!!! Use multiple sources!!!

rainbow146

Kevin你这个故事是写给自己看还是作为作业提交的？写得太有感情了……

对于seminar课程的必读文献，还有一种写summary的形式，就是围绕此次seminar的主题，提出一个问题，然后综合这一组文章中的观点去回答这个问题，这种训练对学术写作的要求更高，对过qualify自然也很有帮助。

举个例子，比如某次seminar讨论某个主题，有若干篇文章，从不同视角、不同方法提出不同的研究问题，那么你可以以“Does methodological diversity contribute to our generation of knowledge in this area or hinder it?”为essay topic，总结这几篇文章。

galluplinc

Kevin你这个故事是写给自己看还是作为作业提交的？写得太有感情了……

对于seminar课程的必读文献，还有一种写summary的形式，就是围绕此次seminar的主题，提出一个问题，然后综合这一组文章中的观点去回答这个问题，这种训练对学术写作的要求更高，对过qualify自然也很有帮助。

举个例子，比如某次seminar讨论某个主题，有若干篇文章，从不同视角、不同方法提出不同的研究问题，那么你可以以“Does methodological diversity contribute to our generation of knowledge in this area or hinder it?”为essay topic，总结这几篇文章。

-- by 会员 rainbow146 (2010/10/4 4:23:34)

没事儿写这个给自己看？？我宁愿出去跑步。。。。


你说的这个也很重要！！但是我觉得这已经不是summary的范畴了，而是PhD们每周必做的功课：thought paper 我们的research methods seminar只要求summary，其他seminar是要求你说的这种thought paper

benxu

Good summary.

It helps with the Comps.

omick

写得真好，受教了。

rainbow146

你们research methods都上成seminar的形式了，牛！

[/quote]



没事儿写这个给自己看？？我宁愿出去跑步。。。。


你说的这个也很重要！！但是我觉得这已经不是summary的范畴了，而是PhD们每周必做的功课：thought paper 我们的research methods seminar只要求summary，其他seminar是要求你说的这种thought paper

-- by 会员 kevinleihuang (2010/10/4 4:56:02)

[/quote]

galluplinc

你们research methods都上成seminar的形式了，牛！

-- by 会员 rainbow146 (2010/10/4 5:16:24)

把research methods上成seminar是悲哀  不是啥好事儿 ........

benxu

同意KEVIN。

非常的同意。。。。。。

lu05721025

。