Atmospheric parameters and chemical compositions for ten stars with metallicities in the region of -2.2< [Fe/H] <-0.6 were precisely determined using high resolution, high signal to noise, spectra. For each star the abundances, for 14 to 27 elements, were derived using both LTE and NLTE approaches. In particular, differences by assuming LTE or NLTE are about 0.10 dex; depending on [Fe/H], Teff, gravity and element lines used in the analysis. We find that the O abundance has the largest error, ranging from 0.10 and 0.2 dex. The best measured elements are Cr, Fe, and Mn; with errors etween 0.03 and 0.11 dex. The stars in our sample were included in previous different observational work. We provide a consistent data analysis. The data dispersion introduced in the literature by different techniques and assumptions used by the different authors is within the observational errors, excepting for HD103095. We compare these results with stellar observations from different data sets and a number of theoretical galactic chemical evolution (GCE) simulations. We find a large scatter in the GCE results, used to study the origin of the elements. Within this scatter as found in previous GCE simulations, we cannot reproduce the evolution of the elemental ratios [Sc/Fe], [Ti/Fe], and [V/Fe] at different metallicities. The stellar yields from core collapse supernovae (CCSN) are likely primarily responsible for this discrepancy. Possible solutions and open problems are discussed.Read less <