In a group, all group members are asked to rate all other group members on a rating scale. This rating scale can have 3 (1..3), 5 (1..5), 7 (1..7) or 9 (1..9) rating points. The rating scale has a neutral mid point (respectively 2, 3, 4 and 5). Application of SSrat::calcgroup calculates a classification into five categories of sociometric status, which is labeled as follows: (1) popular, (2) rejected, (3) negelected, (4) controversial and (5) average.
calcgroup(schoolid = 1, groupid = 1, dataframe, scalelength = c(5, 3, 7,
9), alpha = c(0.1, 0.05, 0.01), NBcriteria = F, printresult = F)
The schoolnumber that identifies the school. Default = 1.
The groupnumber that identifies the group. Default = 1.
The dataframe with the rating data. This dataframe should have columns schoolid, groupid, respid, and for n raters columns r01 to rn, with a maximum of r99. Function readratdatfixed can be used to create such a dataframe from a prepared text file.
Either 3, 5, 7 or 9. Default = 5.
The significance levels to be applied to the probability distributions of the four total scores that have been derived from the ratings. By choosing an appropriate alpha, the user can fine tune the criterion for the status determination. A list of various alphas can be supplied. Default is the list (.10, .05, .01).
A boolean. When TRUE, the classification criteria of
Newcomb & Bukowski (1983) will be applied, in stead of the SSrat criteria.
These criteria are applicable to rating scales of length 3. When this option
is selected with longer scales, the midscore is conversed to 2, alls cores
larger than the midscore are conversed to 3 and all scores lower than the
midscore are conversed to 1. When another recoding scheme of the scores is
preferred, the input ratings should be recoded to 1, 2 and 3 before the use
of this function (use car::recode
).
Boolean which identifies whether the calculated results should be shown. Default is False.
dataframe with the most relevant results, as calculated for each respondent by SSrat
school id as entered
group id as entered
respondent id as entered
number of assessors who have given a valid rating
total rating Sympathy
total rating Antipathy
total rating Preference
total rating Impact
Social Determination as attributed by SSrat, applying alpha = .xx. Defaults to SS.10, SS.05 and SS.01
matrix of Sympathy ratings
matrix of Antipathy ratings
matrix of Preferences
matrix of Impact scores
probability referring to left-sided testing of tr.S
probability referring to right-sided testing of tr.S
expected value of tr.S
probability referring to left-sided testing of tr.A
probability referring to right-sided testing of tr.A
expected value of tr.A
probability referring to left-sided testing of tr.P
probability referring to right-sided testing of tr.P
expected value of tr.P
probability referring to left-sided testing of tr.I
probability referring to right-sided testing of tr.I
expected value of tr.I
It is assumed that the data are gathered on a single (2R+1)-point rating scale. For a 7-point scale, R = 3. The scale midpoint must represent a neutral judgment. The data are arranged in a matrix P, with rows belonging to assessors and columns belonging to the assessed.Let P[i, k] denote the rating given by assessor i to group member k. First P* scores are calculated by subtracting R+1 from the values of P. The Sympathy scores S are calculated by replacing all negative scores of P* by zero. The Antipathy score A are calculated by replacing all positive scores of P* by zero and taking the absolute value. P* = S - A.The Preference score P are equal to the ratings obtained. The Impact scores I are calculated by taking the absolute values of P*. I = S + A. In the next step, sum scores are calculated over columns. Then the distributions of these sumscores are calculated, using the principle of convolution. Lastly, the positions in the distributions are calculated to identify persons with scores in the areas of the lower and higher alpha percent. This allows us to translate the criteria of Coie et al. (1983) into probability terms. A person k obtains the following social determinations (E is expected value): Popular = sum(P[,k]) significantly high, sum(S[,k]) > E(sum(S[,k])) and sum(A[,k]) < E(sum(A[,k])); Rejected = sum(P[,k]) significantly low, sum(S[,k]) < E(sum(S[,k])) and sum(A[,k]) > E(sum(A[,k])); Neglected = sum(I[,k]) significantly low, sum(S[,k]) < E(sum(S[,k])) and sum(A[,k]) < E(sum(A[,k])); Controversial = sum(I[,k]) significantly high, sum(S[,k]) > E(sum(S[,k])) and sum(A[,k]) > E(sum(A[,k])); Average = remaining group members.
When the criteria of Newcomb & Bukowski (1993) are applied, the most liked nominations LM are the ratings > R and the least liked nominations LL are the ratings < R, and the impact score SI = LM + LL. The criteria for a person k are: Popular = sum(LM[,k]) significantly high, sum(LL[,k]) < E(sum(LL[,k])); Rejected = sum(LL[,k]) significantly high, sum(LM[,k]) < E(sum(LM[,k])); Neglected = sum(SI[,k]) significantly low; Controversial = sum(LM[,k]) significantly high and sum(LL[,k] > E(sum(LL[,k])) or sum(LL[,k]) significantly high and sum(LM[,k] > E(sum(LM[,k])); Average = remaining group members.
Coie, J.D., & Dodge, K.A. (1983). Continuities and changes in children's social status: A five-year longitudinal study. Merril-Palmer Quarterly, 29, 261-282. Newcomb, A. F., & Bukowski, W. M. (1983). Social impact and social preference as determinants of children's peer group status. Developmental Psychology, 19, 856-867. Maassen, G. H. and Landsheer, J. A. (1998). SSRAT: The processing of rating scales for the determination of two-dimensional sociometric status. Behavior Research Methods Instruments & Computers, 30(4), 674-679.
# NOT RUN {
data(example5.rat)
# calc SSRAT results fot this group
out =calcgroup (school=1, group=23, data=example5.rat, scalelength="3")
out$dataframe
# calc Newcomb & Bukowski results for this group
out =calcgroup (school=1, group=23, data=example5.rat, scalelength="3", NBcriteria=TRUE)
out$dataframe
# calc Newcomb & Bukowski results for example1
data(example1.rat)
out =calcgroup (school=1, group=1, data=example1.rat, scalelength="7", NBcriteria=TRUE)
out$dataframe
# calc SSrat results for example1
out =calcgroup (school=1, group=1, data=example1.rat, scalelength="7")
out$dataframe
# }
Run the code above in your browser using DataLab