I encountered a perplexing set of results after running xtpcse. R-squared = 1.1892. Any ideas why the R2 is greater than 1.0? I am modeling the the average probability in a county-week that new voters are a person of color.

Directly below, of course, is the stata code. Full results are pasted below, and further below is a plot of observed vs. fitted. When I ran .reg I obtained an R-squared = 0.5284 . It would be nice, but having a value for R2 is not crucial for me to know. I am more concerned about whether there is some other problem that is biasing the estimates of the coefficients and SE.
Code:
. xtpcse poc avr_ph1 avr_ph2 mvr  stg1d stg2d stg3d stg1r stg3r  mdssn pg  srrndrv2p_s  stg2d_x_mvr_mc   mdssn_x_mvr_mc   pg_x_mvr_mc  stg3d_x_avr_ph1_mc  stg3r_x_avr_ph1_mc        stg3d_x_srrndrv2p_smc         wa_x_srrndrv2p_smc  marion_x_srrndrv2p_smc  desch_x_srrndrv2p_smc doug_x_srrndrv2p_smc  coos_x_srrndrv2p_smc   mvrndprvs1yrSTG1_s mvrndprvs1yrSTG2_s mvrndprvs1yrSTG3_s mvrndprvs1yrMDSSN_s mvrndprvs1yrPG_s  poc_cnstnc_ph1 poc_cnstnc_ph2 poc_cnstnc_mvr     _23t30 _31t37  _44t66 _67over      lg_olmoth_acm_presco_s3 lg_acm_mvr_cnt_s3 lg_ph1acm_presco_s3 benton clackamas coos curry deschutes douglas gilliam grant harney hood_river jackson jefferson josephine klamath lane lincoln linn malheur marion morrow multnomah polk umatilla wasco washington yamhill  w1 w2 w3 w4 w5 w6 w7 w9 w10 w11 w14 w15 w16 w17 w19 w25 w27 w28 w29 w34 w39 if (reg_yr_2==8|reg_yr_2==12|reg_yr_2==16)&(  _16cntrfcl_2_x==0)    [aw= AW_pres_mt]
DATA: I tried using datatex but the varlist is about 5 times longer than what the command will accept. So I will verbally describe the data. Data has been collapsed by county and weeks. N=3,9434 county-week combinations. Analytical weights were used. Dependent variable (POC) is the average probability in a county week that a new voter was a person of color. Independent vars include 1) the percentage of new voters who registered by each type of registration source (avr_ph1 avr_ph2 mvr ), 2) the percentage of voters registered during a particular stage of election year (stg1d stg2d stg3d stg1r stg3r mdssn pg), 3) interactions between stages of the election year and type of registration source, 4) other continous variables, 5) dummy variables for counties, and 6) dummy variables for week of the registration season.



Array

Array