Data and Methodology
Data for the study was obtained through the administration of a questionnaire instrument to professional real estate valuers who are members of the Ghana Institution of Surveyors (GhIS) in Accra, Ghana. The data instrument requested the valuers to provide an estimate of the market value for a hypothetical residential property at a particular date based on the same set of instructions. In addition, the respondents were asked to indicate the main sources of property market data they use and the reliability of each source.
The subject matter of the valuation was a leasehold estate in a single story three-bedroom house with an unexpired term of ten years sited on 0.093 ha of land located within the SHC neighborhood along the Liberation Road, Airport Residential Area, Accra, a commonly known area. This property was used because the design is well-known by valuers in the study site and market data for properties in the area where it was located is comparatively easy to come by. Thus, it was to ensure that a lot more valuers could participate in the research. All the property details were given to the respondents. These included the specific location of the property, its construction details, fixtures and fittings, external works, access to services, neighborhood characteristics, total floor area, sketch ground floor and location plans, title and planning/building permit status, and ground rent reserved. Also, information on the method employed by the respondents to undertake the valuation and why they employed the method was obtained.
Prior to the administration, a pre-testing of the questionnaire was undertaken to ensure that it passed the face and content validities test. This process among other things requested four experienced valuers to evaluate the questionnaire in terms of whether or not it covers what it was envisaged to address, and the effectiveness of how the research variables were to be measured. A total of 110 questionnaires were administered to the respondents. The respondents were selected based on purposive and snowball sampling techniques. These sampling techniques were employed due to a lack of a reliable sample frame. Although the GhIS provides a yearly list of valuers in good standing in Ghana, there is no such list specifically for valuers in Accra. Also, the lists do not often have the address and location of valuers. Accordingly, probability sampling could not be undertaken. A response rate of 63.64% was obtained.
The extent of variation in the market valuations obtained from the questionnaire survey was assessed by estimating the coefficient of variation and percentage median deviation of the valuations. Thus, the coefficient of variation of all the reported market values was estimated. The percentage median deviation was used because evaluation of the distribution of the reported market values was not normal based on both Levene and Simonov Kolmogorov tests. The median was, therefore, more representative of the distribution of the values than the mean (Field, 2005). The use of the percentage median variation to report the extent of variation is similar to the study by Adair et al. (1996), which used the percentage mean deviation to do so.
Using absolute values and ignoring negative signs, the percentage median deviation of each reported market value from the median market value was assessed as follows:
PMDj = Percentage median deviation of each reported market value from the median market value;
Xj = Each reported market value; and MD = Median market value.
The second part of the questionnaire instrument requested respondents to indicate how frequently they use different sources as well as to rate the reliability of these sources. A Likert scale was used, and the respondents were requested to rate the reliability of the identified property market data sources on a scale of 1-5 (1 = very unreliable, 2 = unreliable, 3 = quite reliable, 4 = reliable, and 5 = very reliable).
Generally, evaluation of an issue on a Likert scale is often undertaken based on consensus around the mean score. Tastle and Wierman (2007) intimate that such consensus could be assessed as follows:
Cns(x|,Ux) = Consensus around the mean score from the responses (the scores given the mean score);
X = The scores; цх = The mean score;
Xj = Each score; and
dx = The range of X (dx = X„m - Xmn).
However, Tastle, Boasson, and Wierman (2009) on the basis of Equation (2.2) suggest that the evaluation could be undertaken using a preferred target (reference). Thus, //v could be replaced by the preferred target and in such circumstances, dx should be multiplied by 2.
The research sought to evaluate the information obtained on the Likert scale in relative terms, and 5 was the highest score on all the scales. Thus, 5 was used as the preferred target, since a score on an attribute closer to or further away from 5 compared to the other attributes studied determined how highly that particular attribute was evaluated compared to the others. Thus, the following equation was used:
Agr = The level of agreement on evaluation of an attribute, and all other variables are as previously defined.