2007年10月27日星期六

领域和方法――来自ChildTrends

AREAS OF SUBSTANTIVE EXPERIENCE

Economic Security:

Children in Poverty 贫困儿童

Children in Working Poor Families 就业贫困家庭的儿童?

Maternal Employment 母亲的就业状况

Welfare Reform Programs and Children 福利改革计划与儿童

Child Care Subsidies 儿童照顾补贴

Heath:

Child Physical Health 儿童体质健康

Asthma 哮喘

Food Insecurity 食品不安全

Obesity 肥胖

Child Mental Health 儿童精神卫生

Behavior and Social Environment:

Adolescent Sexual Behavior 青少年性行为

Juvenile Delinquency 未成年人犯罪

Youth Risk Behaviors 青少年危险行为

Youth Positive Behavior 青少年积极行为

Education:

School Engagement 入学

Readiness of Schools上学意愿

Family and Population:

Family Economic Well-Being 家庭经济福利

Family Strengths 家庭的力量

Family Violence 家庭暴力

Fatherhood 父亲

Nonresident Father Involvement 父亲缺失家庭?

Marriage and Family 婚姻和家庭

Latino Families 少数民族家庭

Race/Ethnic Relations 民族和种族关系

Immigration 迁移和流动

Intergenerational Ties/Relations 代际关系

Contraceptive Use 避孕方式

Teen Fertility 青少年生育

Nonmarital Fertility 未婚生育

Foster Care/Adoption 儿童收养

Sibling Relationships兄弟姐妹关系

Child Abuse and Neglect 儿童虐待和忽视

Children with Special Needs 特殊需要儿童

Adolescent Peer and Romantic Relationships青少年同伴关系和恋爱关系?

Peer Interactions and Relationships同伴交往和同伴关系

Transition to Adulthood 成年转变

Culture and Children's Development 文化和儿童发展

 

AREAS OF METHODOLOGICAL EXPERTISE

Qualitative Research Methods 定性研究方法

Focus Groups 焦点组座谈 (定性研究方法中的一种)

Literature Review 文献回顾 (元分析meta analysis? )

Survey Research Methods 调查研究方法

Multivariate Analysis 多变量分析

Regression Analysis 回归分析

Structural Equation Modeling 结构方程模型

不熟悉的领域:

Policy Analysis政策分析 (非常重要的能力,研究价值的体现)

Process Evaluation过程评估 (不太明白)

Program Evaluation 项目评估 (评估这个领域发展很快,但是国内未重视)   

Cost and Cost/Benefit Analysis 成本效益分析 (是项目或政策评估的一种方法?)

其他学科的方法论:

Measures Development 测量发展?测量理论

Psychometric Analysis 精神心理分析?

Cognitive Testing 认知测验

Technical Support to Programs 项目技术支持

 

2007年9月27日星期四

A Visual Guide to Stata Graphics 值得记下的几点

 
这书真漂亮,也真难看。
 
漂亮是里面的图做的漂亮,难看也是里面的图难以看明白,太多了!
 
所幸后面的几点还不错,记下:
 
1.Build graphs slowly. Rather than trying to make a final graph all at once, try to build the graph slowly adding one option at a time.(一次一选项)
2.When possible, model graphs from existing examples.(收集一些常用的画图语法作为模板)
 
Stata的画图只能慢慢来了。

2007年9月25日星期二

人民大学图书馆的stata书籍

 
1.现代医学统计方法与 Stata 应用 陈峰主编 (已买)
 
2.A handbook of statistical analyses using Stata / Sophia Rabe-Hesketh, Brian S. Everitt  3rd ed. Chapman & Hall/CRC, c2004. (已有电子版)
 
3.Statistics with Stata : updated for version 9 / Lawrence C. Hamilton  Thomson-Brooks/Cole, c2006. (已复印、电子版)
 
4.A gentle introduction to Stata / Alan C. Acock  StataCorp LP, 2006.
 
5.A visual guide to Stata graphics / Michael N. Mitchell StataCorp LP, c2004. (已有电子版)
   
6.An introduction to modern econometrics using Stata / Christopher F. Baum Stata Press, c2006. (已有电子版)
  
7.An introduction to survival analysis using Stata / Mario A. Cleves, William W. Gould, Roberto G. Gutierrez  Rev. ed. Stata Press, c2004. 
  
8.Data analysis using stata / Ulrich Kohler, Frauke Kreuter  Stata Press, c2005. (已复印)
 
9.Maximum likelihood estimation with stata / William Gould, Jeffrey Pitblado, William Sribney  3rd ed. Stata Press, c2006. (已复印)
 
10.Multilevel and longitudinal modeling using Stata / Sophia Rabe-Hesketh, Anders Skrondal Stata Press, c2005.
 
11.Regression models for categorical dependent variables using Stata / J. Scott Long, Jeremy Freese  2nd ed. StataCorp LP, c2006. (已有电子版)
 
两年之内能把这些书看完就不错了,然后再弄套stata10或以上的手册,啧啧,资料就逼近完美了。

   
 

2007年7月31日星期二

七月流火

 
 
 
留白
点点点

2007年6月15日星期五

SPSS的产品备忘录(数据分析步骤分和软件模块分):

Don't ask if SPSS can do this or that! Ask how it can be done!——Raynald Levesque
 
更丰富的Client/Server功能,支持客户端自由数据和用户评分界面;更强大的数据管理功能,包括克隆数据集、在单个SPSS进程中打开多重数据集、长数值标签等;更强大的编程功能,SPSS语法中嵌入了源代码开放的Python程序;更强大的统计分析功能,包括偏好尺度、数据校验过程、在多变量中识别异常值、增强的Trends分析、回归插补,随机回归插补和多重贝叶斯插补,以及更完善的结构方程建模分析;更优秀的图表功能,使用Chart Bulider用户界面,支持SPSS公司的图形化编程语言图形生成语言 (GPL)

数据分析步骤分:
节约时间和金钱: Sample Power, SPSS Complex Samples™, and SPSS Conjoint
收集/输入数据--Web: Dimensions and SPSS Data Entry
轻松快捷的数据获取: SPSS® Base
有效的数据管理和准备: SPSS Base, SPSS Missing Value Analysis, SPSS Complex Samples, SPSS Text Analysis for Surveys, and Dimensions
使用各种各样的统计分析过程, 针对更精确的模型来分析数据: SPSS Base, SPSS Regression Models, SPSS Advanced Models, SPSS Complex Samples, SPSS Data Validation, SPSS Classification Trees, SPSS Categories, SPSS Exact Tests, SPSS Trends, AnswerTree, and Amos
清楚地报告你的结果给需要的人: SPSS Base, SPSS Tables, SPSS Maps, and Dimensions
在网络上、出版物或产品展示中, 与他人共享结果: SmartViewer Web Server and SmartScore

软件模块分:
SPSS Advanced---用高级工具包分析复杂关系:提供了一组成熟的单变量和多变量分析技术来解决现实问题,它能方便地建立更灵活、更成熟的模型,在处理嵌套数据时得到更精确的预测模型。它可用于分析事件历史和持续时间数据,还可自定义工具,用内建的宏程序库进一步地定制工具组,以便扩展分析功能用于更专门的用途。Advanced Models广泛应用于医学研究、制造业、制药、市场研究。Advanced可帮助您理解顾客的行业表现、增加交叉销售额、提高客户忠诚度、赢得新的商机。
SPSS Base---高质量的数据分析工作帮您做出更好的决策:具有强大的统计分析功能。由于取消了一般分析工具普遍存在的文件大小限制,因此它可以处理其它分析工具无法解决的大型数据,能够更快地访问和分析大型数据资料,帮助您快捷轻松地为分析作数据准备而无需了解SQL。SPSS Base用全面的技术进行数据分析,包括了多种最受欢迎的基本统计功能,您还可以把分析结果回写到数据库。数据分析完毕后,SPSS Base将结果用在线分析处理的方式展示,并用交互式图表清晰地表达分析结果。
SPSS Categories---帮您揭示分类数据中的独特关系:优秀的对应分析程序,用启发性的二维图和感知图让您清晰地洞察数据中的关系,使您可以更完整和方便地分析数据。通过类似传统的回归分析、主成份分析及典型相关分析的分析方法,帮您处理和了解分类数据及定序数据。SPSS Categories 的尺度优化过程让您不再受限于二维的表格,并使用图表可视化地展示多变量分类数据中的关系,充分表现出动态、交互的分析结果,让您从分类数据中得到更丰富的信息。
SPSS Classification Trees---轻松确定分组和预测结果:基于数据挖掘中发展起来的树结构模型对分类变量或连续变量进行预测,可以方便、快速的对样本进行细分。使用Classification Trees还可建立决策树来确认分组并预测结果,利用直觉式的树形图,颜色分类图,和表格协助研究人员轻松确认和评估区隔。它提供四种强大的决策树算法(CHAID、Exhaustive CHAID、CART和QUEST),提供训练数据及测试数据的模型评估比较,提供Gain、Response、Index、Profit、ROI图,帮助评估风险及效益。
SPSS Complex Sample---从调查数据中获得更准确的结果:绝大多数统计软件假定数据来自简单随机抽样,但如果您进行的是大规模的调查,往往使用的是通过较复杂的抽样得到的数据。SPSS Complex Sample拥有专门的规划工具和统计方法,能减少得出错误或误导性推论的风险。它提供各种向导来制定抽样方案或详细定义样本,并提供专门的技术来解决样本设计问题以及标准误差的复杂计算,使您能够正确、简单地从分层、整群或者多阶段抽样数据来计算出各种统计量及它们的标准误。SPSS Complex Sample将抽样设计融入调查分析之中,使您对复杂抽样数据的总体得到更加有效的统计推论,对于调查、市场、民意研究人员或者社会科学家来说是必不可少的统计工具。
SPSS Conjoint---揭示消费者偏好,帮您改进产品、有效定价:包含三个相互关联过程的一个系统,用于进行全轮廓联合分析。联合分析使研究人员了解消费者的偏好,或在一定产品属性及其水平条件下进行产品评定。联合分析考虑应包括的产品属性以及各属性水平的组合,通过正交设计生成一个包含适量属性组合的正交主效果片段因子设计。SPSS Conjoint帮助市场研究人员和新产品开发部门了解在消费者心目中哪些产品属性及其组合是重要的,了解他们最偏爱的属性水平是什么,从而能够进行新产品研究、定价研究、品牌研究,并预测各种潜在新产品的市场份额。 
SPSS Data Validation ---改善数据准备,获取更多准确结果:所有的研究者都需要为分析而准备数据,这往往是个耗费时间和冗长乏味的手工操作。使用SPSS Data Validation的附加模块,您可以轻松地鉴别出可疑的或者无效的记录、变量和数据值;观察缺失值的模式;概括变量的分布;从而判定数据的有效性并做出相应地处理。SPSS Data Validation能够根据数据的不同测量水平来使用指定的规则进行数据检查,并可以迅速地找到多元的奇异值。这种流水化的数据准备过程让您能够更快的为分析做好准备,得到更精确的结果。
SPSS Exact Test---在小样本或小子集的情况下得出正确结论:提供世界顶级的精确检验和推论功能,传统情况下,要得到更好正确的结论,每一单元需要有五个以上数据,SPSS Exact Tests 为您解除了这种限制。Exact Tests帮助研究人员不管数据情况如何总是可以找到合适的统计检验方法进行小规模研究,而结果仍然是正确的和可信的。SPSS Exact Tests拥有超过30个精确检验,涵盖了小型或大型数据集所有的非参数检验和定类数据处理的问题,您无需学习任何新的统计理论和过程,就可以用和传统统计方法一样的方式来解释精确检验的分析结果。Exact Tests可以保留您原先设计的分类或自然分类,让您在不破坏分类的情况下继续进行分析,从而消除使用传统检验造成的疑虑。
SPSS Maps---强大的集成化地图分析,精确地描绘您的战略方针:集成化的强大的地图程序,它把SPSS数据转换成地理信息,使您的地区数据和地理因素可视化,用各种图表和地图选项给您展现一个清晰的图像,表示地理因素是如何影响业务的。研究人员可以使用Maps建立专业的地图、描绘战略方针,用SPSS Maps Server处理繁重的地图绘制工作。SPSS Maps可以帮助分析地理因素和业务之间的关联性、自动解释销售数据、显示销售趋势等,它方便地把您的数据和全球样本及数据进行比较,可以帮您确定季节性、客户动态、竞争情况,等等。SPSS Maps包括众多的免费样本数据,还有40 多个全球MapInfo 数据文件。
SPSS Missing Value Analysis---排除数据中隐含的偏差,得出更精确的结论:缺失数据会带来偏差或错误的分析结果,简单代入法或者简单的回归法都不能正确地填补缺失值,SPSS Missing Values Analysis帮助研究人员在分析过程中排除数据中隐含的偏差,得出更精确的结论。Missing Values Analysis用六种灵活的诊断报告来评估缺失值是否会影响分析结论,更好地了解它们的特性。它通过快捷地诊断缺失值,得到更精确的摘要统计量,方便地用估计值替换缺失值,得到精确的结论。
SPSS Programmability---延伸统计的深度及广度,并可连贯操作程序以达自动化:可以让您使用外部语言来执行SPSS一连串的分析动作以达到自动化的目的。您可藉由撰写内嵌在SPSS 14.0里的Python程序来控制SPSS的各式语法工作执行,如执行设定变量属性、观察程序输出、错误码或条件状态等。运用外部程序结合新的后端API处理,可扩大语法执行工作的弹性。透过条件执行式(IF/Then/Else)、循环控制式(For/While),这些几乎都会出现在外部语言的叙述,套用在新增的SPSS Programmability模块上,使得控制语法工作的程序更有 效率、更方便。
SPSS Regression---用强大的回归模块做出更好的预示:提供大量的非线性建模工具、多维尺度帮助研究人员预测行为和态度,这些模型要好于用简单的线性回归所建立的模型。它将分析从数据约束中解放出来,提供多种步进方式来选择重要的、最好地预示了响应变量的连续或分类变量,建立可控制的模型及表达式进行非线性模型的参数估计,还使用新的多项式逻辑斯谛回归来预示多于两个类别的分类结果。
SPSS Tables---用包括所有统计量、易于理解的表格来展现分析结果:提供35种单元和摘要统计量,能够更方便地显示多重序列数据,它能串接所有的维度,以在同一表格中显示包含不同统计量的各种变量。Tables用更深入的分析,轻松地处理复选题与缺失值,用包括所有统计量、易于理解的表格来展现分析结果。
SPSS Trends---用强有力的时间序列分析工具做更好的预测: 目前功能最强的时间序列分析工具,是分析历史资料、建立模型与预测未来事件的强有力的工具,能帮助研究人员做更好的预测。Trends提供一流的评估过程,用简单、有效的平滑技术进行高质量的预测,使用SPSS Trends可完成多种任务,包括生产管理、数据处理、预算管理、公共政策研究等。
SPSS Amos6.0---使用结构方程式,探索变量间的关系:通过建立结构方程式模型,用标准方法以及在此基础上扩展的方法进行多元分析,比普通最小二乘回归和探索性因子分析更进一步,能获得更精确、丰富的综合分析结果。使用Amos直观的拖放式绘图工具,您可以快速地以路径图定制模型而无需编程。在有缺失值的情况下,Amos使用Full Information Maximum Likelihood方法仍然可以自动计算正确的标准误及适当的统计量,降低估算值偏差。 最新版本Amos6.0还增加了探索性结果方程模型、辅助多组分析、高级文本输出、扩展的Amos编程环境等功能。Amos被广泛地应用于顾客满意度分析等领域。
Answer Tree---用决策树更直观、有效地定位顾客群:功能强大的决策树软件,能够轻松快速地发现细分区域,建立决策树模型,并发现隐藏的趋势。从分析、过滤、到专业的输出,Answer Tree都能为研究人员带来巨大的方便。通过四大决策树计算法则--CHAID、Exhaustive CHAID、Classification and Regression Tree、Quest,Answer Tree为您提供无以匹敌的 分析。在SPSS for windows中,SPSS Base和Classification Trees结合使用,也能达到Answer Tree的功能。
Dimensions---通过理解客户提高业绩:支持整个调查研究流程的一整套产品,包含从项目设计、数据收集到高级分析和报告的过程。同时,Dimensions还是一个平台,不仅可以构建单个调查,而且可以构建整个研究应用。在任何与客户互动的形式中,您可以通过调查获得客户反馈,并且把调查结果与其他形式的业务信息整合起来。

2007年6月7日星期四

SPSS和Stata到底能处理多大的数据?

 
五一放假期间把SAS 9.1.3 下载下来,也安装完了,由于缺少完美的SID和使用SAS的经验,很快就从机器里删除了。其实还是挺舍不得它强大的数据处理和商业应用的能力,以后再说吧。还是把手头上SPSS和Stata练熟才好。前一段时间因为要处理一个上亿个case的数据,没有仔细想就把SPSS和Stata放弃了,潜意识就认为它们不能处理或很难处理这么大的数据。后来VB都用上了才算搞定。但,SPSS和Stata到底能处理多大的数据?答案如下:
 
对于SPSS,10.0版本就能处理【2^15 = 32,768 variables  ;2^31 = 2,147,483,648 cases】。10.0及以后的SPSS的variables和cases数量都变为2^31或2^31-1了。其他人的说明:The below is by Jon Peck of SPSS, Inc., and applies to all recent versions of SPSS.There are several points to making regarding very wide files and huge datasets.

        2. The overhead of reading and writing extremely wide cases when you are doubtless not using more than a small fraction of them will limit performance.  And you don't want to be paging the variable dictionary.  If you have lots of RAM, you can probably reach between 32,000 and 100,000 variables before memory paging degrades performance seriously.(我也不会用到那么多的variables)

        5. These points apply mainly to the number of variables.  The number of cases is not subject to the same problems, because the cases are not generally all mapped into memory by SPSS (although Windows may cache them).  However, there are some procedures that because of their computational requirements do have to hold the entire dataset in memory, so those would not scale well up to immense numbers of cases.(估计用SPSS对上亿个cases的数据做个频数分布都非常苦难)

        Modern database practice would be to break up your variables into cohesive subsets and combine these with join (MATCH FILES in SPSS) operations when you need variables from more than one subset.  SPSS is not a relational database, but working this way will be much more efficient and practical with very large numbers of variables. (大量的variables能用subsets,大量的cases呢)

对于Stata,一句话,靠可用的内存。具体见FAQs by Kevin S. Turner, StataCorp:

      1.Under all current 32-bit Windows operating systems (Windows 95, 98, ME, NT, 2000, XP, Vista), the total available address space for any application is 2.1 GB. If you have a dataset larger than 2.1 GB, you will not be able to load it on Stata for Windows.

      2.Unfortunately, even if your dataset is under the 2.1 GB limit, you may run into difficulty when loading it into Stata. The fault again lies with how Windows manages the 2.1 GB address space. You may be surprised to find that a 1.4 GB dataset loaded fine one time, but failed to load a subsequent time. This is simply an unfortunate side effect of Windows memory management.
 
     3.The 64-bit platform will enable you to work with very large datasets. Depending on your operating system, you should be able to allocate as much memory as you have on the machine minus the system requirements. To take advantage of this technology, you will need 64-bit compatible hardware, a 64-bit operating system, and, of course, a 64-bit version of Stata.

     4.As a last resort, you may consider trimming any unnecessary data from your dataset or dividing the dataset into two files. Depending on your data and analysis this may not be feasible, and is only offered as a suggestion.

2007年6月6日星期三

Is Sex Selection of Births Undesirable?

 
作者:加里贝克尔
 
点评:居然看到了他的博客,还在上面找到了这篇文章!出生性别选择是坏事吗?这种提法由贝克尔提出来尤其显得意味深长。简言之,  由于生男孩有利,通过父母的选择行为,出生性别比的上升导致了女孩数量少于男孩,较少的女孩可以获得更多的资源,这样成年后,女性会因为稀缺而在社会中获得更多的优势,男性的价值会相对下降,长期而言,必然会导致女孩偏好,而失衡的出生性别比会再次通过父母选择得到修正。虽然现在中国确实是处于女性稀缺的婚姻挤压状态下,但是"男尊女卑"的中国传统文化岂能轻易改变。受过高等教育的女性较高的不婚率说明了她们宁愿不结婚,也不愿意去走出"男不高攀,女不低就"的桎梏。

In China in 2005, 118 boys were born for every 100 girls born. This ratio is far above the normal biological ratio of about 106 boys to 100 girls. The sexual disparity in China has resulted from a combination of low birth rates, a preference in China for boys when parents only have one or two children, and the spread of ultrasound techniques in that country that allow the sex of fetuses to be identified and then aborted if parents do not like the sex. Similar trends have emerged in India and South Korea as well.

More sophisticated and expensive methods permit parents to raise their chances of a male baby even before a woman becomes pregnant. Considered most reliable is a method that involves in vitro fertilization, drugs to stimulate the mother's ovaries, surgery, and other steps. The total cost can exceed $20,000, so this method clearly is only available to richer persons.

Are there good reasons to object to sex selection, either by abortion or more sophisticated methods? On Feb. 1 the Committee on Ethics of the American College of Obstetricians and Gynecologists (the ACOG) did issue an opinion objecting on the grounds that it is unethical for physicians to participate in sex selection by parents that was based not on potential for sex-linked genetic disorders, but solely on family balancing of personal preferences. This opinion about the ethics of sexual selection applied "regardless of the timing of the selection ( i.e., preconception or post conception) or the stage of development of the embryo or fetus".

Such an opinion seems strange in light of the general support by physicians and the Supreme Court of abortions by parents "solely" to satisfy their personal preferences about timing or number of children. What is so different about sex-selected abortions that would lead the ACOG with its over 51,000 members who provide health care to women to oppose abortions to satisfy parental desires for additional boys or girls while supporting the general right to abortion? The ACOG tries to provide an answer by claiming that sex selection through any method may "ultimately support sexist practices."

It is not clear what the ACOG means by sexist practices, but all the evidence on sexual preferences in the United States and other richer countries indicates an overwhelming desire for variety-boys and girls- rather than a strong preference for either sex. So sex-selected abortions in these countries is unlikely to have much of an effect on the overall sex ratio, although it would affect the distribution of boys and girls in different families.

I concentrate my remaining discussion on the implications of sex-selected abortions in countries where it raises the number of boys relative to girls. China, South Korea, and other countries have tried to implement control over sex selection by making it illegal to use ultrasound techniques to select the sex of children. However, these regulations are notoriously difficult to implement since doctors may say "congratulations" when an ultrasound test reveals a boy, and remain silent when the fetus is a girl.

Abortions of girl fetuses would reduce average family size if parents who prefer boys would end up with larger families than they would like because they cannot control the sex of their offspring. The effect on family size could go the other way, however, if the fear of having girls discourages parents from having additional children. These effects on family size could be important, but I ignore them in the following discussion and concentrate on the effects of a lower number of girl babies relative to boys compared to the biological natural girl-boy ratio of a little below 50-50.

One might expect parents who abort fetuses of sexes they do not want to treat their children better than they would otherwise since they now are satisfied with the sexes of their children. In such cases, sex-selected abortions against girls would improve rather than worsen the average treatment of girls since parents would be happier with the girls they have than if they had girls who were not really wanted. It is no surprise, for example, that orphanages in China predominantly have girls (and some handicapped boys), given the preference for boys in the traditional Chinese culture.

What about the overall effects in a society of skewing the sex ratio of births toward boys? The fewer girls who are born presumably would be better off since they would be better educated, and in other ways better treated by parents who want them. This would be reinforced if the effect of sex selected abortions is to lower the overall birth rate since it is well established that families with fewer children invest more in each one, girls as well as boys.

As children become adults in cohorts with a high ratio of boys, the advantage of girls and women increases since they are scarcer. It is claimed that young women in China are already at a premium as potential mates because strong sex-selection has been going on ever since the one child policy was introduced in the early 1980's. Prior to the spread of ultrasound techniques, sex selection occurred through sending girls to orphanages, neglect, and in some case even engaging in female infanticide.

To be sure, if the value of girls as wives and girlfriends, and in other ways, rises because they are scarcer, then the value of boys as husbands and boyfriends tends to fall. However, it is not apparent why that should call for policies that prevent sex-selected techniques, unless the interests of men were motivating these policies. To use an analogy, a shift of demand in an economy toward services and away from manufacturing because of a shift in "preferences" toward services- as has occurred in the United States and other rich countries- benefits women relative to men since women are more likely to work in services than are men. Yet no one would claim that society should prevent such preferences because they help (indirectly) one sex over another.

The great statistician and biologist, R. A. Fisher, used a celebrated biological analysis to explain why the sex ratio remains close to 50-50 in non-human species. An economic analysis based on incentives gives results that are related to Fisher's result. An improvement in the position of women due to a decline in the number of girls relative to boys leads to some correction in the sex ratio as parental choices respond in the long run to the more favorable position of girls. If women are in greater demand as wives and in the economy when they are in scarcer supply, some parents will decide that having girls has advantages, possibly through receiving generous bride prices when daughters marry. This would shift "preferences" toward having girls. The long run outcome would not necessarily be the biological natural ratio of a little more boy births than girl births, but it should be closer to that ratio than the current ratios in some Asian countries.