关于 R 中未加权 ANES 调查的问题

Question about unweighted ANES survey in R

这可能更像是一个统计问题。我正在尝试使用 2016 年 ANES 调查来理解 R 中未加权的频率。我使用 anesr 下载了数据,并且正在使用 Anthony Damico 关于 lodown 的教科书:http://asdfree.com/american-national-election-study-anes.html

ANES 指南:https://electionstudies.org/wp-content/uploads/2018/12/anes_timeseries_2016_userguidecodebook.pdf

ANES 是一项小组调查,包含选举前采访和 post 选举采访。

抱歉,我无法 post 基础数据。但本质上,我试图理解一个问题。

假设我想获得对 V161195x 的回复的未加权计数,它总结了关于应如何处理非法带到美国的儿童的观点。使用未加权的数据框,我 运行:

table(anes_df$V161195x)

Yielding: 
-9   -8    1    2    3    4    5    6 
  47   29  329  332  112  435 1437 1549 

-9和-8拒绝回答或N/A.

然后我使用 survey 包进行复杂的样本调查设计和更易于阅读的输出。

anes_design <- 
  svydesign( 
    ~V160202, # full sample weight 
    strata = ~V160201, 
    data = anes_df , 
    weights = ~V160102 , # full sample weight 
    nest = TRUE 
  )
  
anes_design <- 
  update( 
    anes_design,
    
    one = 1, # dummy 1 for each record

    undoc_kids =
      factor(V161195x , levels = 1:6 , labels =
                c( 'should sent back - favor a great deal' ,
                   'should sent back - favor a moderate amount' ,
                   'should sent back - favor a little' ,
                   'should allow to stay - favor a little' ,
                   'should allow to stay - favor a moderate amount' ,
                   'should allow to stay - favor a great deal' )
      )     
  )


svyby( ~ one , ~ undoc_kids , anes_design , unwtd.count )

# yields 
should sent back - favor a great deal                   should sent back - favor a great deal    270  0
should sent back - favor a moderate amount         should sent back - favor a moderate amount    292  0
should sent back - favor a little                           should sent back - favor a little    102  0
should allow to stay - favor a little                   should allow to stay - favor a little    361  0
should allow to stay - favor a moderate amount should allow to stay - favor a moderate amount   1216  0
should allow to stay - favor a great deal           should allow to stay - favor a great deal   1348  0


这与第一个 table 中的输出不匹配,即使它们都未加权。似乎 svyby 函数只包括在选举前和 post 选举调查中的观察结果,但问题只在选举前调查中。

当我使用 V160101(仅限选举前)而不是 V160202 重新 运行 分析时,输出匹配...那么最好使用哪个?

因为您使用的是 post-选举权重变量,对于不在 post-选举调查中的人来说,该变量缺失(或可能为零),您只会得到观察结果在两项调查中。