Skip to contents

SPM linear-regression t-field

Source: R/stats-spm-regress.R
spmRegression.Rd

Computes an SPM{t} field testing, at every node, the slope of a linear regression of the waveform on a continuous predictor. The node-wise statistic is the ordinary least-squares slope t-value (identical to summary(lm(y ~ predictor)) at that node), and field-level significance uses the same random-field-theory threshold as spmTTest() with df = n - 2.

Usage

spmRegression(x, predictor, alpha = 0.05, two_tailed = TRUE)

Arguments

x

A PhysioExperiment or a numeric matrix (time x observations).

predictor

Numeric covariate, one value per observation (column).

alpha

Significance level (default 0.05).

two_tailed

Logical; two-tailed slope test (default TRUE).

Value

A list of class "spm_result" (test_type = "regression") with the t field, RFT threshold, significant clusters, pointwise p_values, df, fwhm, resel_count, and the fitted slope/intercept fields.

References

Pataky 2016; Friston et al. 2007 (RFT). spm1d.stats.regress.

See also

spmTTest(), spmMANOVA(), spmSnPM()

Examples

set.seed(1)
pred <- rnorm(20)
data <- matrix(rnorm(100 * 20), nrow = 100)
data[40:60, ] <- data[40:60, ] + outer(rep(1.5, 21), pred)  # slope 30% window
spmRegression(data, pred)
#> SPM Analysis Result
#> ==================
#> Test type: regression
#> Time points: 100
#> Alpha: 0.050
#> Threshold: 4.374
#> FWHM (smoothness): 2.24
#> Resel count: 44.12
#> 
#> Significant clusters: 4
#>   Cluster 1: [40-40] extent=1, p=0.0114
#>   Cluster 2: [42-44] extent=3, p=0.0000
#>   Cluster 3: [46-57] extent=12, p=0.0000
#>   Cluster 4: [59-60] extent=2, p=0.0001