SSRA {SSRA}

Sakai Sequential Relation Analysis

Description

This function conducts Sakai Sequential Relation Analysis (SSRA) based on Sakai 2016.

Usage

SSRA(dat, r.crt = 0.3, mu.sq = 0, mu.eq = Inf, d.sq = 0.2, d.eq = 0.2,
  pairwise = TRUE, method = c("pearson", "kendall", "spearman"), alpha = 0.05,
  p.adjust.method = c("holm", "hochberg", "hommel",
                      "bonferroni", "BH", "BY", "fdr", "none"),
  digits = 3, vnames = TRUE, order = c("no", "decreasing", "increasing"),
  exclude = TRUE, output = TRUE)

Arguments

dat

requires a data frame with polytomous data

r.crt

correlation coefficient criterion to be judged 'sequential' or 'equivalent

mu.sq

Absolute mean difference criterion to be judged 'sequential'

mu.eq

maximal absolute mean difference to be judged 'equivalent'

d.sq

effect size for mean difference criterion to be judged 'sequential'

d.eq

maximal effect size Cohen's d to be judged 'equivalent'

pairwise

pairwise deletion of missing data, if pairwise = FALSE listwise deletion is applied

method

character string indicating which correlation coefficient to be used, 'pearson' = Pearson's product moment correlation coefficien 'spearman' = Spearman's rho statistic 'kendall' = Kendall's tau (default)

alpha

significance level

p.adjust.method

p-value correction method for multiple comparisons, see: ?p.adjust (default = holm)

digits

integer indicating the number of decimal places to be used

vnames

use variable names for labeling?

order

sort by item mean of j and k?

exclude

exclude paths with no relationship?

output

print result table?

Details

In Sakai Sequential Relation Analysis (SSRA), a pair of items is judged 'sequential', if there is a higher correlation and a bigger mean difference than defined criterions between the two items. If there is a higher correlation and a smaller mean difference than defined criterions between the two items, the relation of the two items is judged 'equal'.

Value

Returns an object of class ssra, to be used for the seqtable function. The object is a list with following entries: 'dat' (data frame), 'call" (function call), 'args' (specification of arguments), 'time' (time of analysis), 'R' (R version), 'package' (package version), and 'restab' (result table). The 'restab' entry has following entries:

j	item j
k	item k
n	sample size
j.mean	mean of item j
j.sd	standard deviation of item j
k.mean	mean of item k
k.sd	standard deviation of item k
r	correlation coefficient
r.t	test statistic of the statistical significanc test for the correlation coefficient
r.p	statistical significance value of the correlation
r.sig	statistical significance of the correlation (0 = not significant / 1 = significant)
r.crt	correlation criterion for judging 'sequential' or 'equal': 'r.p < alpha' and 'r > r.crt' (0 = no / 1 = yes)
m.diff	mean difference
sd.diff	standard deviation difference
m.diff.eff	effect size Cohen's d for dependent samples
m.t	test statistic of the statistical significanc test for mean difference
m.p	statistical significance value of the mean difference
m.sig	statistical significance of the mean difference (0 = not significant / 1 = significant)
m.crt.sq	mean difference criteria for judging 'sequential': 'm.diff.p < alpha', 'm.diff > mu.sq' and 'm.diff.eff > d.sq' (0 = no / -1 = yes negative / 1 = yes postive)
m.crt.eq	mean difference criteria for judging 'equivalence': statistical significant and 'm <= mu.eq' 'd <= d.sq' (0 = no 1 = yes)
seq	sequential relation of item pairs ("+","-", "")
eq	equivalence of item pairs ("=" or "")
order	order structure of item pairs ("=", "+","-")

Author(s)

Takuya Yanagida [e-mail], Keiko Sakai [e-mail]

References

Takeya, M. (1991). A new test theory: Structural analyses for educational information. Tokyo: Waseda University Press.

See Also

seqtable, TSSA, plot.ssra, scatterplot

Examples

# Example data based on Takeya (1991)

# Sakai Sequential Relation Analysis
# ordering assesed according to the correlation coefficient and mean difference
SSRA(exdat)