test_that(
  "evidence ratio is as expected",
  {
    # from Motulsky & Christopolous (2004), p. 146
    # if the AICc scores differ by 5.0, then the evidence ratio equals 12.18.
    expect_equal(calc_evidence_ratio(0.0, 5.0), 12.18, tolerance = 0.01)
    expect_equal(calc_evidence_ratio(10.0, 15.0), 12.18, tolerance = 0.01)

    # If the difference in AICc scores equals 10, then the evidence ratio is 148
    expect_equal(calc_evidence_ratio(0.0, 10.0), 148, tolerance = 1)
    expect_equal(calc_evidence_ratio(50.0, 60.0), 148, tolerance = 1)
  }
)

test_that(
  "comparison is as expected",
  {
    # mock up some dummy fit results
    fit_a <- list(aic = 10.0)
    fit_b <- list(aic = 5.0)

    fits <- list(a = fit_a, b = fit_b)

    comparison <- compare_fits(fits = fits)

    # first row should be the preferred model
    expect_true(comparison[1, "preferred"])
    expect_false(comparison[2, "preferred"])

    expect_equal(comparison$aic, c(5, 10))

    expect_equal(comparison$preferred_rel_fit_delta_aic, c(0, -5))

    expect_equal(
      comparison$preferred_rel_fit_evidence_ratio,
      c(1, calc_evidence_ratio(5, 10))
    )
  }
)