We study the radio properties at 1.4 GHz of Seyfert galaxies with strong forbidden highionization lines (FHILs), selected from the Sloan Digital Sky Survey – a large-sized sample containing nearly equal proportion of diverse range of Seyfert galaxies showing similar redshift distributions compiled by Gelbord et al. (2009) using the Very Large Array survey images. The radio detection rate is low, 49%, which is lower than the detection rate of several other known Seyfert galaxy samples. These galaxies show low star formation rates and the radio emission is dominated by the active nucleus with ≤10% contribution from thermal emission, and possibly, none show evidence for relativistic beaming. The radio detection rate, distributions of radio power, and correlations between radio power and line luminosities or X-ray luminosity for narrow-line Seyfert 1 (NLS1), Seyfert 1 and Seyfert 2 galaxies are consistent with the predictions of the unified scheme hypothesis. Using correlation between radio and [O III] λ 5007 Å luminosities, we show that ∼8% sample sources are radio-intermediate and the remaining are radio-quiet. There is possibly an ionization stratification associated with clouds on scales of 0.1–1.0 kpc, which have large optical depths at 1.4GHz, and it seems these clouds are responsible for free–free absorption of radio emission from the core; hence, leading to low radio detection rate for these FHIL-emitting Seyfert galaxies. Spiral galaxies, having bright star-like nuclei covering a wide range of ionization, are known as Seyfert galaxies. They are generally classified into type 1 and type 2, which depends only on relative nuclear emission-line widths (Khachikian & Weedman 1974) . The Seyfert 2 galaxies have relatively narrow permitted hydrogen lines and narrow forbidden lines, whereas Seyfert 1 galaxies have broad permitted hydrogen lines and narrow forbidden lines. The widths of narrow lines and of broad lines in terms of full width at half maximum (FWHM) are ≈ 300–1,000 km s ⁻¹ and ≥ 1,000 km s ⁻¹ for type 2 and type 1, respectively. An orientation based unified scheme is often used in explaining the classification of a Seyfert galaxy, where the Seyfert 2 represent edge-on source and the Seyfert 1 galaxy is its pole-on equivalent (e.g., Antonucci 1993 ; Urry & Padovani 1995 ; Deo et al. 2007 ; Netzer 2015) . Observational evidence for and against this unified scheme hypothesis exist on various scales and in all wavebands (see Singh et al. 2013 , for a brief summary). In all these studies, sample selection is the key; for example, biases against obscured sources, and biases towards dusty sources are suggested for the Optical-/UVselected samples; whereas biases towards sources with higher level of nuclear star formation are suggested for the infrared-selected sources (Schmitt et al. 2001 ; Mason et al. 2012) . Similarly, optical-/UV-/X-ray selected flux limited samples, are all likely to have intrinsically more luminous Seyfert 2 galaxies than Seyfert 1 galaxies (Heckman et al. 2005 ; Wang et al. 2009) . In short, issues pertaining to sample selection are key and new efforts to test the unified scheme hypothesis with superior and well-chosen samples continue to be made (Lal et al. 2011 ; Urry & Padovani 1995 , and see also Section 2.2). This paper tests the validity and limitations of the unified scheme hypothesis for Seyfert galaxies. Here, we present results for a Seyfert galaxy sample defined by Gelbord et al. (2009 , hereafter GMW09), using Very Large Array (VLA) survey images. This large sample contains a diverse types of Seyfert galaxies. Below, we first explain classifications of Seyfert galaxies (Section 2.1) and give a brief description of the sample (Section 2.2). We use the radio maps from Faint Images of the Radio Sky at Twenty–Centimeters (FIRST; Becker et al. 1995 ) and NRAO VLA Sky Survey (NVSS; Condon et al. 1998 ) for the sample objects (Section 3), and interpret our results (Section 4) and their implications on the unified scheme hypothesis (Section 5). Finally, we summarize our conclusions in Section 6. Following GMW09, we also adopt the same cosmological parameters from Wilkinson Microwave Anisotropy Probe . Hence, distance-dependent quantities are calculated assuming H ₀ = 71 km s ⁻¹ Mpc ⁻¹ , Ω _m = 0.27, and Ω _Λ = 0.73 (Spergel et al. 2003) . When archive radio data is available at other frequencies, we determine radio spectral index, α which we define in the sense that S_ν ∝ ν^α , where S_ν is the flux density and ν is the frequency. All coordinates mentioned below are for J2000 epoch. In addition to Seyfert galaxies of type 1 and type 2; Osterbrock (1981) and Osterbrock & Pogge (1985) introduced additional fractional classifications; these appear to have a mix of properties from both types, as the broad component of H β becomes weaker as compared to the narrower component, the Seyfert type changes from 1.2 → 1.5 → 1.8 → 1.9. Seyfert 1.2 and 1.5 galaxies have composite spectra, with both broad and narrow components easily recognizable and the former have broad component being stronger than narrow component, whereas the latter have both components that are comparable. In Seyfert 1.8 and 1.9 galaxies, in addition to the narrow components, very weak but recognizable broad components of H β and H α , or of H α alone, respectively are present (Osterbrock 1981) . In terms of the unified scheme hypothesis these intermediate Seyfert types are thought to lie at a varying angles between pole-on and edge-on views, with Seyfert 1.2 galaxies being close to pole-on, whereas Seyfert 1.9 galaxies being close to edge-on. Osterbrock & Pogge (1985) proposed yet another type of Seyfert galaxies, the narrow-line Seyfert 1 (NLS1) galaxies (Valencia et al. 2012) . These objects have permitted line-widths much smaller than typical Seyfert 1 galaxies. However, they differ from Seyfert 2 galaxies, in the sense that optical spectra of NLS1 galaxies show several characteristics normally associated with Seyfert 1 galaxies, such as [O III]/H β ratios of less than 3, permitted lines are broader than forbidden lines, and blends of lines such as Fe II or [Fe VII] or [FeX]. Again, in the unified scheme frame-work, NLS1 galaxies could be sources with views very close to pole-on (Urry & Padovani 1995) . Our goal is to study the radio properties of Seyfert galaxies and their implications on the unified scheme hypothesis, and we use Seyfert galaxies listed in GMW09, with strong forbidden high-ionization line (FHIL) emission. This sample contains a diverse types of Seyfert galaxies, including NLS1 galaxies. It has Almost all known nearby, z ≲ 0.1 Seyfert galaxy samples, e.g., the bright Seyfert galaxies sample (Giuricin et. al. 1990) , the CfA Seyfert sample (Huchra & Burg 1992) , the 12 μ m Seyfert sample (Rush et al. 1993 ; Hunt & Malkan 1999) , the far-infrared selected Seyfert galaxy sample (Roy et al. 1994) , the matched Seyfert sample (Lal et al. 2011) , etc. rarely have NLS1 galaxies in them and do not have appropriate proportion of Seyfert types discussed above. Other FHIL-emitting samples of Seyfert galaxies in the literature (e.g., De Robertis & Osterbrock 1984 ; Erkens et al. 1997 ; Murayama & Taniguchi 1998 ; Veilleux 1998; Nagao et al. 2000) are all relatively small and heterogeneous. Instead, the GMW09 sample of Seyfert galaxies with strong FHILs is one of the largest containing nearly equal proportion of diverse Seyfert types and hence, by far the most homogeneous to date. In order to discuss differences between NLS1, Seyfert 1 and Seyfert 2 galaxies and its implications on the unified scheme hypothesis, we henceforth address Seyfert 1.0 and 1.5, and Seyfert 1.9 and 2.0 as Seyfert 1, and Seyfert 2 galaxies, respectively. Note that GMW09 found one galaxy with an unusual spectrum that is not Seyfert-like (object 45, spectral type “gal” in GMW09), which was neither included by GMW09 nor it is included in the rest of this paper. We also do not include radio galaxy 3C234 (object 22, spectral type “Seyfert 1.9”), though it was included by GMW09, because it is a radio-loud active galactic nucleus (AGN; Laing et al. 1983 ). One another source (object 53, spectral type “NLS1”) is not covered by the FIRST survey and is not detected in NVSS image; this too is excluded. Our final sample consists of 61 (from parent list of 64) Seyfert galaxies. All these 61 objects have measured [FeX], [FeVII] and [O III] fluxes and among these 59% (36/61) objects have measured [FeXI] fluxes, which is twice as many as compared to earlier samples in the literature. The mean redshifts and their dispersions for individual sub-classes of Seyfert galaxies are, These means are based on redshifts defined from the observed wavelengths of the [S II] doublet. It seems that the Seyfert 1.0 and 1.5 galaxies in the sample are farther set of objects with respect to the mean redshift for the sample 0.0869 ± 0.0532 possibly due to the bias mentioned above. However, much of it is due to three outlier objects, ID-12 (spectral type 1.0), and ID-16 and ID-28 (spectral type 1.5), which are only objects with z > 0.2 in the sample; and excluding these objects, the means are 0.0840± 0.0426 and 0.0843± 0.0391 for Seyfert 1.0 and Seyfert 1.5, respectively. Barring this, the mean redshifts and their distributions (see Figure 4 , GMW09) for diverse types of Seyfert galaxies agree among themselves. In addition, the median redshifts of all Seyfert types also agree. Therefore, apart from only possible bias, bias against Seyfert 1 galaxies with the broadest permitted lines and any FHIL-emitting sources dominated by lines with lower ionization potentials, presence of which is unclear (GMW09), these Seyfert sub-classes do not differ in redshift distribution. We, thus use this sample containing diverse types of Seyfert galaxies with FHIL features to study the radio properties and its implications on the unified scheme hypothesis. Our earlier effort of investigating radio properties of type 2 SDSS quasars (Lal & Ho 2010) yielded a detection rate of 59% (35/59) at 8.4 GHz (X-band) survey. This detection rate was essentially identical to that obtained from FIRST survey images at 1.4 GHz with 5′′ resolution, which observed 56 out of the 59 sources in our sample and detected 35 (63%), even though the sensitivity of FIRST survey is nearly an order of magnitude lower than that of our X-band survey (Lal & Ho 2010) . Furthermore, the sensitivity of NVSS survey images at 1.4 GHz with 45′′ resolution is ∼0.45 mJy beam ⁻¹ (Condon et al. 1998) , a factor of three lower than the FIRST survey images and surprisingly the detection rate was again identical. Since, the sample of 61 Seyfert galaxies with FHIL-emitting features also come from SDSS, we believe that deeper new observations may not be necessary to understand basic radio properties. We therefore use FIRST survey images (at 1.4GHz, ∼5′′ resolution) together with NVSS images (at 1.4GHz, ∼45′′ resolution), and archive VLA data, when available along with published optical and X-ray (GMW09) data extensively to draw our conclusions. Galaxies were considered detected if the peak flux density noise, where the noise of each map was determined from a source–free region. In total 30 sources from the sample of 61 were detected, of which eight were resolved. The radio images of eight sources showing extended, resolved structures, are shown in Figure 1 . These source images are arranged in the order of increasing R.A., and their restoring beams (with position angles = 0°), root-mean-squared (r.m.s.) values and contour levels in the maps are given in Table 1 . The ratio of the peak surface brightness and the r.m.s. noise, or the dynamic range for the FIRST survey, and the NVSS images are between ∼5 and 66, and ∼6 and 43, respectively. For the undetected sources, the flux density or its upper limit is equal to 5 × r.m.s. The sizes (major and minor axes) and elongation direction (position angle) of the detected components/sources were determined by fitting two-dimensional Gaussians to each detected source and deconvolving those Gaussians from the synthesized beams using Astronomical Image Processing System (AIPS 1 ) task JMFIT. The data for the sources with complex morphologies, the flux densities were found from integration over the maps covering the source using AIPS task TVSTAT. A summary of derived radio parameters from the FIRST survey images, along with optical parameters is given in Table 2 , together with Seyfert galaxy types in the following order: NLS1, Seyfert 1.0, 1.5, 1.9, and 2.0, which are further arranged in the order of increasing R.A. The individual columns are: (1) Object name; (2) ID Number as defined by GMW09; (3) spectral classification; (4) redshift from the observed wavelength of the [S II] doublet; (5) luminosity distance 2 ; (6) r.m.s. noise in the map; (7) peak flux density; (8) integrated flux density; (9) deconvolved (half maximum) size of the source (major and minor axes); (10) largest linear size (l.l.s.); (11) comments on the source structure; and (12) integrated flux density as determined from the NVSS image. For each object in Table 2 , we have classified the radio structure into following categories (Ulvestad & Wilson 1984 ; Ulvestad & Ho 2001) : “U” (unresolved), “S” (slightly resolved), “L” (linear, usually core–jet or double), and “E” (extended); where slightly resolved and extended sources are those whose deconvolved size is ≥ 0.5 times the synthesized beam width (Ulvestad & Ho 2001 ; Lal & Ho 2010) . Whenever we obtain zero arcsec as the deconvolved size in any one dimension, we ascribe “U” (unresolved) status to the object. There are 8 out of the 30 confirmed detections which are extended. SDSS J082930.59+081238.1 — This source appears extended in the FIRST image and is classified as diffuse with faint extended emission. The low-surface brightness feature toward the northeast seen at 1.4 GHz is possibly a jet. SDSS J092343.00+225432.6 — The radio core is unresolved but the source appears to have east–west elongation. The low-surface brightness contours suggest the source to be amorphous and is classified as an extended morphology. SDSS J110704.52+320630.0 — A marginal case of core-jet (Kharb et al. 2010) morphology. The faint jet is seen as emanating from the core along the northern direction. SDSS J115704.84+524903.7 — The source is resolved, with typical core–jet morphology. SDSS J122930.41+384620.7 — The source is marginally resolved with an extension along east. SDSS J134607.71+332210.8 — The low-surface brightness contours at 3 σ –5 σ levels suggest that the source is possibly a diffuse, with extended structure. Assuming the structure at the phase-center and the extended feature adjacent to it on the east to be real, the source is likely to be of core–jet morphology. SDSS J153552.40+575409.5 — This object is also classified as Mrk 290 (PG1534+580) at z = 0.0296 (NED) in the literature. The radio core is slightly resolved at 1.4 GHz. An image using archive VLA A -array configuration data at 5.0GHz (project-code AF0402, observe-date 2003-Jun-17) shows the source to be unresolved. The integrated spectral index using FIRST survey image and this archive data, = −0.73±0.04, typical of many other Seyfert galaxies (Lal et al. 2011) . SDSS J220233.85−073225.0 — This source has a double-lobed extended morphology at 1.4 GHz. It is the southeast radio lobe, which is associated with the optical position. Hence, it is possible that radio core is slightly resolved showing a core–jet morphology. Of the 61 FHIL-emitting Seyfert galaxies in the sample, 30 (49%) have been detected down to the FIRST survey r.m.s. sensitivity limit of 0.15 mJy beam ⁻¹ (Becker et al. 1995) . This detection rate further lowers, 34% (21/61) if we instead consider NVSS data. Furthermore, this detection rate is: Note that barring type 2 SDSS quasar sample (Lal & Ho 2010) , all other Seyfert galaxy samples are nearby (redshift ≲ 0.03), and except radio observations of two Seyfert galaxy samples, Roy et al. (1994) and (Panessa & Giroletti 2013) , which are at higher angular resolutions, the radio data of all other samples are at similar, arcsecond-scale resolutions. Among these nearby Seyfert galaxy samples, the detection rates of FHIL-emitting Seyfert galaxy and far-infrared selected samples are similar. Looking more closely only at the detected sources, the sample size of 21 objects; namely, Thus, the detection rates of NLS1, Seyfert 1 and Seyfert 2 galaxies are 0.25 ± 0.16, 0.27 ± 0.11, and 0.50±0.07, respectively. Alternatively, the radio detection rate of compact radio structure for NLS1, Seyfert 1 and Seyfert 2 is consistent with the unified scheme hypothesis. Ulvestad et al. (1981) , Whittle et al. (1986) , Evans et al. (1991) and Falcke et al. (1998) noted that compact radio emission is closely associated with individual narrow-line region clouds, and Roy et al. (1994 , and see also Norris et al. 1992) first invoked a model that free–free absorption by these clouds may explain the low detection rate in FHIL-emitting Seyfert galaxies. In particular, when observing Seyfert galaxies showing very low radio emission from the nucleus, either (i) the narrow-line region or (ii) the individual narrow-line region clouds would absorb radio emission; i.e., if the optical depths are above unity due to free–free absorption then either the narrow-line region or the individual narrow-line clouds will block our view of the core and we would not detect it in radio band. Although this model involve optical depth effects in the narrow-line region, its geometry and filling factor, it reconciles our low radio detection rate. Furthermore, in this model, the narrow-line region clouds would become optically thin at higher frequencies and optically thick at further lower frequencies (Norris et al. 1992) . So new radio observations at a sufficiently high frequencies should find higher detection rates and similarly, at sufficiently low frequencies should find lower detection rates, which is a possible test if free–free absorption by these clouds is indeed responsible for the low detection rate at 1.4GHz for the FHIL-emitting Seyfert galaxies in the sample. Twenty-two of 30 (∼73%) detected objects are classified as “U” or “S”, unresolved or slightly resolved and the radio emission in these is largely from the central active nucleus. This result shows that a significant fraction of Seyfert galaxies do not have double/triple radio sources, instead they have unresolved structures on a scale of several hundred parsecs to a few kiloparsecs. The double radio sources that have been found in Seyfert galaxies generally have angular separations smaller than the resolution of the present data (Kukula et al. 1995 ; Thean et al. 2000 ; Lal et al. 2004) . Although the exact number must await high-resolution radio maps of the complete FHIL-emitting Seyfert sample, eight (out of detected 30 = 27%) sources ( Figure 1 ) have marginally extended, about a few kiloparsec scale to a few tens of kiloparsec scale structures. The largest projected linear sizes (l.l.s.), determined from largest-angular-sizes (l.a.s.) listed in Table 1 , for eight extended sources shown in Figure 1 corresponding to the 3 σ contour level of the 1.4GHz maps are listed in Table 2 . Clearly Seyfert 1 and NLS1 galaxies have larger range in projected linear sizes as compared to Seyfert 2 galaxies. Unfortunately with only eight sources, three Seyfert 1, three NLS1 and two Seyfert 2, this is small number statistics. The statistical test 3 gives a poor significance level of 0.25 that the distribution of projected linear sizes of NLS1, Seyfert 1 and Seyfert 2 galaxies are same. Higher-resolution observations are therefore needed to resolve the FIRST survey images into structures with physical sizes less than a kiloparsec, as often seen in nearby Seyfert galaxies (e.g., Kukula et al. 1995 ; Thean et al. 2000 ; Lal et al. 2004) . These resolved morphologies would provide improved statistics and test the distribution of projected linear sizes at a higher significance level. Figure 2 shows the distribution of the radio luminosity detected on 5′′-scales (FIRST images) and 45′′-scales (NVSS images) for all Seyfert sub-classes. Radio luminosities for our sample sources are K -corrected, restframe values determined from observed flux densities at 1.4GHz and corresponding luminosity distances (Lal & Ho 2010) . Since majority (22 of the 30 confirmed detections) of the sources are unresolved and correspond to core emission, we assume a typical spectral index, α = +0.5 from cores of Seyfert galaxies (Kellermann & Owen 1988) . We have also chosen the radio luminosity bin widths sufficiently larger than the r.m.s. noise in the maps to account for upper limits for undetected sources, which are not explicitly denoted in Figure 2 . Although non-detections lie in the luminosity-bins corresponding to the low end of radio powers in both panels, the distribution of radio powers at 1.4 GHz is continuous, showing no separation between the detected and the non-detected objects. Furthermore, changing the radio luminosity bin size does not make the distribution bimodal (or multimodal). This suggests that the detected and undetected sources from the FHIL-emitting Seyfert galaxy sample irrespective of Seyfert class denote a single population as a whole. In addition, the radio power distributions are not significant different between NLS1 galaxies, Seyfert 1 galaxies and Seyfert 2 galaxies. The K–S test gives a significance level of 0.026 or better that the distribution of Seyfert 1 along with NLS1 galaxies and Seyfert 2 galaxies are same for the FIRST and NVSS data. Thus, we conclude that the radio power distributions of Seyfert galaxy types are similar and are consistent with the expectations of the unified scheme. Baum & Heckman (1989) and Rawlings & Saunders (1991) showed that a correlation exists between the radio luminosity at 5GHz and the [O III] λ 5007Å narrowline luminosity. Figure 3 illustrates the distribution of integrated radio power (at 5 GHz) versus [O III] luminosity for the large, heterogeneous sample of AGNs compiled by Xu et al. (1999) , after correcting for the cosmology. The data for FHIL-emitting Seyfert galaxies come from 1.4 GHz; and to be consistent with Xu et al. (1999) , we converted these measurements to 5 GHz assuming a flat spectrum spectral index of α = 0 and determine the monochromatic power. These monochromatic powers change at the most by a factor of 12%, if instead we assume a slightly inverted or steep spectra (see also Section 5.1.3) and it does not change our interpretations below. Note that we have computed [O III] luminosities using combined fluxes of [O III] core and [O III] wing components (supplementary material: Table A4, Column 8; GMW09). It is clear that AGNs separate into the two families of radio-loud and radioquiet AGNs with a significant gap between them; i.e., the radio luminosities are different by a factor of 10 ³ –10 ⁴ between the two groups at a given [O III] luminosity. All the objects in FHIL-emitting Seyfert galaxy sample are of radio-quiet nature, with ∼8% that lie in between the two families of radio-loud and radio-quiet AGNs, also called radio-intermediate AGNs. It is not surprising that all the non-detected sources are radio-quiet. Linear fits to the FHIL-emitting Seyfert galaxy sample excluding the radio-intermediate objects yield which is consistent within error bars with the linear fit by Xu et al. (excluding the radio-intermediate objects; 1999) obtained for the radio-quiet AGNs. Furthermore, assuming 50% of the flux density detected in the FIRST survey images is also detected at milliarcsec-scales (Lal et al. 2011) , the inferred brightness temperature, T _B ≥ 5 × 10 ⁸ K for our sample objects, which is again typical of Seyfert galaxies (Broderick & Fender 2011) . In radio galaxies, one-sided structures are often associated with relativistic jets and the jet to counter-jet ratios are used to obtain quantitative estimates of relativistic beaming. Here, unfortunately a large fraction of sources are non-detected (∼51% in FIRST and ∼67% in NVSS survey images) and a majority of the rest of the detected sources are unresolved (22 objects out of 30). The FIRST survey images probes structures on scales smaller than the NVSS survey images (see Section 3). Therefore, we assume any difference of the emissions between NVSS and FIRST images as extended emission, which is not Doppler boosted; and instead, if beaming is present, the emission detected in FIRST image would possibly be Doppler boosted (Lal et al. 2011) . Hence, the ratio of the possibly boosted and the extended radio flux densities, or the R -parameter could be used here to investigate relativistic beaming. Since a majority of sources are not detected, we focus only on the detected sources (see also Section 5.1.1). The ratio of flux density for these detected (20 of 61) sources in both, FIRST and NVSS survey images is close to 1.0 with mean and median being 1.06±0.04 and 1.02, respectively. Clearly, much higher resolution, scales of milli-arcsecond would be required to address relativistic beaming. Additionally, the use of R -parameter as measure of relativistic beaming comes with a caveat. Lal et al. (2011) have shown that Seyfert galaxies show radio variability and the FIRST and the NVSS data for the FHIL-emitting Seyfert sample are not simultaneous. Therefore, here we have made an assumption that these sample objects do not show radio variability (see also Mundell et al. 2009) . Additionally, Falcke et al. (1996) showed that Lorentz factors of γ = 2–4 in radio-quiet AGNs are sufficient to boost the radio emission into the radio-intermediate regime or into the radio-loud regime. Five objects from FHIL-emitting Seyfert galaxy sample are of radio-intermediate kind: Although this is a small number statistics, majority are non-type 1 Seyfert galaxies, and it is well known that Seyfert 2 galaxies being edge-on counterparts of beamed population, which are not expected to show relativistic beaming. Therefore, it is possible that FHIL-emitting Seyfert galaxy sample are radio-quiet and there is no evidence of relativistic beaming consistent with the predictions of unified scheme hypothesis, which is further supported by other Seyfert galaxy samples (Lal et al. 2011) . In Figure 4 , we plot correlations between radio power and luminosities of [FeX] (top-left panel), [FeXI] (top-right panel), [FeVII] (middle-left panel), [O III] (middle-right panel), [O I] (bottom-left panel), and H _α (bottom-right panel. [O III] and H _α line luminosities correspond to combined core and wing components and combined narrow and broad H _α model components, respectively (GMW09). In Section 2.1, we discussed inherent biases in the sample and hence possible difference between NLS1, Seyfert 1 and Seyfert 2 galaxies are seen in H _α luminosity, X-ray luminosity, and ratio of [FeX] and [FeVII] lines. Barring, correlation between radio power and H _α luminosity, X-ray luminosity and ratio of [FeX] and [FeVII] lines, which we discuss below (Section 5.2.1, 5.2.2, 5.2.3, respectively), none suggest that the distribution of NLS1, Seyfert 1 and Seyfert 2 galaxies are dissimilar. We therefore find no observational evidence against the unified scheme hypothesis. The K–S test gives a significance level of 0.031 or better that the Seyfert types, Seyfert 1 galaxies along with NLS1 galaxies and Seyfert 2 galaxies are drawn from the same parent population, and there is no significant differences between Seyfert 1 galaxies along with NLS1 galaxies and Seyfert 2 galaxies, based on the distributions of [FeX], [FeXI], [FeVII], [O III], and [O I], which was also independently concluded by GMW09. Below we discuss those cases where we clearly see segregation between the Seyfert types and their implications on the unified scheme hypothesis. The radio luminosity versus H _α luminosity correlation plot in Figure 4 (bottom-right panel) show that, both types of Seyfert galaxies have similar radio powers, but Seyfert 2 galaxies have systematically lower H _α luminosity than Seyfert 1 galaxies along with NLS1 galaxies. the latter is because the broad component of H _α in Seyfert 2 galaxies is obscured by the obscuring torus and is always not detected, which is more evident in Figure A2b (from supplementary material) of GMW09. Since the sample is selected on FHIL emission, lines with ionization potential ≳ 100 eV (GMW09), their radio emission is expected to have similar distributions of radio powers within the framework of unified scheme hypothesis. This is indeed the case, i.e., both types of Seyfert galaxies have similar radio powers, consistent with the predictions of unified scheme. There are only 32 sample sources with ROSAT detections, of which only three are Seyfert 2 galaxies, which on average have lower X-ray luminosities but their radio powers are similar to rest of the sample sources. In Figure 5 , we show the correlation between radio and soft X-ray luminosities fitted to Seyfert galaxy populations. Unfortunately, Chandra/XMM-Newton data do not exist for the sample. Seyfert 1 along with NLS1 galaxies and Seyfert 2 galaxies are known to have, on average, steeper soft X-ray spectra and flatter hard X-ray spectra, respectively (Kruper et al. 1990 ; Mas-Hesse et al. 1994 ; Boller 2000 ; Dadina 2008) . Here, we used single uniform photon index, Γ = 1.5 as well as steeper photon index (Γ = 3.0) data from GMW09 for Seyfert 1 along with NLS1 galaxies to convert ROSAT count rates to fluxes (GMW09; Cappi et al. 2006 ; Dadina 2008) . We find that the same correlation function applies between radio and soft X-ray luminosities for all types of Seyfert galaxies with no systematic differences between them. We therefore conclude that there is no difference between radio luminosities of NLS1, Seyfert 1 and Seyfert 2 galaxies, consistent with unified scheme. Differences between the two types of Seyfert galaxies for the ratio of [FeX] and [FeVII] provide evidence that in Seyfert 2 galaxies there exist partially obscured [FeX] emission and hence low ratio of these two lines (GMW09). This is consistent within the stratified wind model framework discussed below (Section 5.3), that the [FeX] is emitted on size scales that is comparable to the size of the dusty molecular torus, and hence obscuration may be important. As shown in Figure 6 , NLS1 galaxies, Seyfert 1 galaxies and Seyfert 2 galaxies show difference in the ratio of [FeX] and [FeVII] which are possibly because of (i) absorption of the [FeX] or (ii) different ionizing continuum (GMW09), where the latter is perhaps due to the bias in the sample. Briefly, in Seyfert 1 galaxies, the line width of the [FeX] line profiles is broader, and the ratios of [FeX] and [FeVII] tend to be higher, whereas it is in opposite sense for the Seyfert 2 galaxies (GMW09). These facts suggest that [FeX]-emitting clouds lie close to the radio core and in certain lines of sights, some fraction of the [FeX] flux (i.e., the broad component) may be hidden by the obscuring torus (GMW09). Hence, there is possibly an excess of ionizing photons from active nucleus and we expect an over-production of [FeX] or the higher ratio of [FeX] and [FeVII] in the NLS1 galaxies and also in the Seyfert 1 galaxies, whereas this would again be in opposite sense for Seyfert 2 galaxies due to the obscured view of active nucleus by the obscuring torus. These are in line with the expectations of the unified scheme hypothesis. Hence, in Figure 6 , we find all types of Seyfert galaxies to have similar radio powers, but the Seyfert 2 galaxies on average have less [FeX] power per unit of [FeVII] power than the Seyfert 1 galaxies or NLS1 galaxies, which is again consistent with the predictions of unified scheme. We first investigate if the radio emission from 8% radio-intermediate sources in the sample been enhanced due to intense star formation. Quantitatively, the mean radio power of radio-intermediate sources, ≃ 8.5×10 ³⁰ erg s ⁻¹ Hz ⁻¹ implying a star formation rate (SFR) of ∼ 220 M _⊙ yr ⁻¹ for stars massive enough to form supernovae, i.e., M ≥ 5 M _⊙ . Mainieri et al. (2005) ; Vignali et al. (2009) and Rosario et al. (2012) showed that such high rates of star formation are seen in high- z AGNs, but they are more than the SFR inferred for (i) typical nearby Seyfert galaxies (Alonso-Herrero et al. 2013) by an order of magnitude (Lal et al. 2011) , and also for (ii) the most luminous AGNs in the SDSS (Liu et al. 2009) . Therefore, it seems that the enhanced radio emission in radio-intermediate sources as compared to radio-quiet sources in the sample is unlikely to be due to intense star formation. Secondly, the radio spectra of most Seyfert galaxies are known to be steep power laws near 1.4GHz, < α > ∼0.8 (Lal et al. 2011) indicating that the radio flux densities at this frequency are dominated by nonthermal emission. However, the strong optical emission lines imply the presence of ionized gas that could contribute significant thermal emission at frequencies ≥1.4 GHz. It is thus important to compare the expected thermal emission and the total radio flux densities observed from Seyfert galaxies in the sample. We now test and demonstrate that both, radio-quiet and radio-intermediate Seyfert galaxies in the sample have low-levels of expected star formation rates, lower than the limits that is needed to account for the observed radio emission. Following Ho (2005 , 2008) and also Lal & Ho (2010) , the [O II]-derived SFRs are estimated to be ∼0.2–10 M _⊙ yr ⁻¹ , with a mean (≈ median) value of ∼0.6 M _⊙ yr ⁻¹ for the Seyfert galaxies in the sample; we use the ratios of ionisation potentials between [O I], [O II] and [O III] emission lines and [O I] and [O III] fluxes, and thereby determine [O II] fluxes. Using Equation 6 of Bell (2003) , the predicted radio emission from SFR is ∼ 1.1×10 ²⁸ erg s ⁻¹ Hz ⁻¹ at 1.4 GHz, which is at least an order of magnitude less than the mean radio luminosity of the sample sources ( Figure 3 and 4 ). Alternatively, assuming all lines of sight through the broad line region are optically thick in the radio, an upper limit to the thermal emission at the mean distance of ∼400 Mpc for objects in the sample gives a flux density of ≲ 0.004 mJy, quite undetectable for typical values of densities of protons and electrons, n _p = n _e ≃ 10 ⁹ cm ⁻³ in broad line clouds, of size ≃ 0.1 pc having a temperature ≃ 1.5 ×10 ⁴ K (Koski 1978) , a filling factor ≃ 10 ⁻⁴ (Ulvestad et al. 1981) , and the H β luminosity equal to the [O III] λ 5007Å luminosity of 10 ⁴² erg s ⁻¹ (Ulvestad et al. 1981) . On similar lines we calculate radio power ≲ 10 ²⁷ erg s ⁻¹ Hz ⁻¹ , using canonical n _e ≃ 10 ³ cm ⁻³ , temperature ≃ 1.5 ×10 ⁴ K, filling factor ≃ a few times 10 ⁻² for narrow line clouds of size ≃ 0.1– 1.0 kpc at mean distance for objects in the sample. Clearly, it is the narrow line region that has relatively larger thermal emission contribution to the observed radio emission as compared to the broad line region. But the derived thermal contributions at 1.4GHz are small, at a level of 10% or less of the observed flux density for objects in the FHIL-emitting Seyfert galaxy sample. Also, Gu et al. (2006) used the observed emission-line ratios along with population synthesis models and concluded that although there is a contribution of starbursts to the nuclear emission in a radio-quiet Seyfert galaxies, the contribution of the hidden active nucleus always dominates. Their conclusion compares well for the FHIL-emitting sources studied here. Finally, in the light of a stratified wind model, first proposed by Osterbrock (1991) , outflowing photoionized clouds produce [FeXI], [FeX], [FeVII] and possibly [O III] (GMW09), with high-ionization lines are produced preferentially at small distances from the core, while low-ionization lines are preferentially produced at larger radii. More specifically [FeXI]- and [FeX]-emitting clouds lie closest to the radio core and are on the scale of broad-line region or of obscuring torus, [FeVII]-emitting clouds are more extended and may approach the scale of narrow-line region, and finally [O III]- and [O II]-emitting clouds, which do not show a velocity shift with respect to [S II] (GMW09), lie farthest from the radio core and more likely are the decelerated part of the stratified wind outflow model (Colbert et al. 1996a , b) . If this model is correct, we expect to find a strong correlation between emission-line shifts and ionization potential of the line-emitting species; implying density and ionization stratification (Komossa et al. 2008) . Such correlations were indeed reported by GMW09 for some of the ion species. Thus, there is a ionization stratification associated with clouds, and possibly clouds on scales of narrow-line region, which have relatively the least amount of ionization stratification, have large optical depths at 1.4GHz. These clouds are responsible for free–free absorption of radio emission from the core, leading to inferred low radio detection rate irrespective of Seyfert types, which is consistent with orientation based unified scheme hypothesis. This paper presents radio properties of a sample of 61 sources containing a diverse range of Seyfert galaxies. This is one of the largest, homogeneous samples of Seyfert galaxies with strong FHIL emission selected from the SDSS to date (GMW09). Here we used highresolution FIRST data (5′′ resolution images, Becker et al. 1995) along with low-resolution NVSS radio data (45′′ resolution images, Condon et al. 1998) , both at 1.4 GHz. Our main results are: However, a weakness remains with the data; these survey images are sensitive to extended emission from starburst activity around the nucleus in addition to the compact emission with the Seyfert core. Future deeper, high-resolution and high-sensitivity radio observations at both, high and low radio frequencies are needed to (i) test and understand the low detection rate, (ii) resolve the FIRST survey images into structures of sizes less than a kiloparsec, (iii) test the predictions of free–free absorption of radio core emission by the narrow-line region clouds, and (iv) understand if these objects have AGN-like, high-brightness temperature flat-spectrum cores, or these have steep-spectrum diffuse emission.