McSpatial (version 2.0)

matchqreg: Sample quantiles and means over time for a matched sample data set

Description

Calculates and graphs sample means and quantiles over time. Intended for but not limited to a data set constructed with matchprop or matchmahal

Usage

matchqreg(form,taumat=c(.10,.25,.50,.75,.90), qreglwr.smooth=TRUE,  window=.50,bandwidth=0,kern="tcub", alldata=FALSE, graph.yhat=TRUE,graph.mean=TRUE,data)

Arguments

form

A formula of the type y~x, where x represents time.

taumat

Vector of quantiles. Default: taumat=c(.10, .25, .50, .75, .90).

qreglwr.smooth

If qreglwr.smooth=T, uses qreglwr to smooth the quantile series. If qreglwr.smooth=F, calculates period by period quantiles.

window

Window size to be passed to qreglwr if qreglwr.smooth=T. Default: 0.50.

bandwidth

Bandwidth to be passed to qreglwr if qreglwr.smooth=T. Default: 0, i.e., not used.

kern

Kernel weighting function to be passed to qreglwr if qreglwr.smooth=T. Default is the tri-cube. Options include "rect", "tria", "epan", "bisq", "tcub", "trwt", and "gauss".

alldata

Indicates how the alldata option should be treated for qreglwr if qreglwr.smooth=T. Default: alldata=F

graph.yhat

If graph.yhat=T, graphs the series of quantile lines. Default: graph.yhat=T.

graph.mean

If graph.mean=T, graphs the means over time. Default: graph.yhat=T.

data

A data frame containing the data. Default: use data in the current working directory.

Value

yhat

Matrix of quantiles for y; actual quantiles if qreglwr.smooth=F and smoothed values if qreglowr.smooth=T. Rows represent time periods and columns represent quantiles.

ymean

Average value of y for each time period.

timevect

Vector of target quantile values.

Details

Calculates means and quantiles of y for each time period present in the variable on the right hand side of the model formula. The quantiles can be varied with the taumat option. If qreglwr.smooth=T, matchqreg uses the qreglwr command to smooth the quantile lines and stores the results in the matrix yhat. The unsmoothed, actual quantile values are stored in yhat if qreglwr.smooth=F. The window, bandwidth, kern, and alldata options are passed on to qreglwr if qreglwr.smooth=T.

Although matchqreg is meant to follow the matchprop or matchmahal command, it can be applied to any data set.

References

Deng, Yongheng, Sing Tien Foo, and Daniel P. McMillen, "Private Residential Price Indices in Singapore," Regional Science and Urban Economics, 42 (2012), 485-494.

Ho, D., Imai, K., King, G, Stuart, E., "Matching as Nonparametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference," Political Analysis 15 (2007), 199-236.

Ho, D., Imai, K., King, G, Stuart, E., "MatchIt: Nonparametric preprocessing for parametric causal inference," Journal of Statistical Software 42 (2011), 1-28..

McMillen, Daniel P., "Repeat Sales as a Matching Estimator," Real Estate Economics 40 (2012), 743-771.

See Also

matchmahal

matchprop

qreglwr

Examples

set.seed(189)
n = 500
# sale dates range from 0-10
# mean and variance of x increase over time, from 1 to 2
# price index for y increases from 0 to 1
timesale <- array(0,dim=n)
x <- rnorm(n,0,1)
for (j in seq(1,10)) {
  timesale <- c(timesale, array(j, dim=n))
  x <- c(x, rnorm(n,j/10,1+j/10))
}
n = length(x)
y <- x*1 + timesale/10 + rnorm(n, 0, sd(x)/2)
fit <- lm(y~x+factor(timesale))
summary(fit)
heddata <- data.frame(y,x,timesale)
summary(heddata)

par(ask=TRUE)
matchdata <- matchprop(timesale~x,data=heddata,ytreat=0,
  distance="logit",discard="both")
table(matchdata$timesale)
fit <- matchqreg(y~timesale,qreglwr.smooth=FALSE, 
  graph.yhat=TRUE,graph.mean=TRUE,data=matchdata)