2.7. Statistical analyses

The data were expressed as meanąS.D. for four animals per group. Statistical
analyses were performed by one-way analysis of variance (ANOVA) followed by
Dunnett's post-test using SPSS (student version 7.5, SPSS Inc., Surrey, UK).
We used the significant level alpha=0.05.

2.6. Statistical analysis

Statistical analyses were performed with Stata/SE 10.1 software
(StataCorp LP, College Station, TX). The results were expressed as the
meanąstandard deviation (SD). Skewness/kurtosis tests for normality
showed that the data obtained for CI-ATP per milliunit of CS and CIATP
per milligram of protein were normally distributed and for
mitochondrial density, CII-ATP per milliunit of CS, and CII-ATP per
milligram of protein after natural logarithm transformation. Differences
between groups were then evaluated on the basis of Student's ttest
or one-way analysis of variance where appropriate. When
significant F values were obtained, individual group means were
tested for differences according to Bonferroni's a posteriori correction
for multiple comparisons. The criterion for significance was P<0.05
for all comparisons.

Linear regression analyses of CI-ATP and CII-ATP per milliunit of CS
on age indicated a marginal (R2=0.20) but significant (P<0.005) age
effect that was lost after patient 9 of the LHON-- group (86 years of
age) was omitted from the database. Therefore, that patient was not
taken into further account.

To examine if early- and late LHON+ differ in their ability to
compensate per cell for the impaired mitochondrial-dependent ATP
synthesis, a Kruskall-Wallis equality-of-populations rank test and a
two-sample Kolmogorov-Smirnov test for equality-of-distribution
functions were performed after assigning the distinction "positive" to
values of CI-ATP per milligram of protein that were equal or higher
than the lowest control value and else "negative".

Statistical analysis

The prevalence rates are proportions of affirmative
responses of all responses. The subjects were divided
into three equal-sized age groups. Differences in pain
prevalence between age groups were assessed by the
Mann-Whitney U-test and the differences between
occupational groups by the chi-square test.

To test whether pain is likely to affect multiple body
sites in some individuals, the number of subjects expected
to have 0, 1, 2, 3, 4, 5, 6, or 7 numbers of sites
with pain was calculated by Poisson's distribution. The
presence of the pain in different anatomical sites in a
subject was assumed independent from the presence of
pain in other sites. The distribution parameter used to
generate the expected number of subjects was the
average number of sites with pain per individual. The
observed frequencies were compared with the expected
frequencies using the chi-square test.

Prevalence ratios (PR) for pain in one anatomical
site relative to another were calculated using Cox's
proportional hazards regression (Lee and Chia 1993;
Lee 1994; Thompson et al. 1998). To reduce the chance
of false positive findings, the Bonferroni correction was
applied. The adjustment for multiple tests was applied
when the significance of deviation of PR from 1.0 was
assessed. The significance level was set at 0.001, adjusted
for 42 multiple tests. To obtain all combinations
of the seven symptoms, or the three groups of symptoms,
each of the dichotomous (0, 1) symptoms (or the
symptom categories) was multiplied with a unique
power of ten, after which the variables were summed.
All analyses were performed using the SPSS Version
12.0.1.

Data Analysis

Data are represented as meanąSEM, n represents the
number of experiences. Statistical analyses were performed
by a one-way analysis of variance and Mann-
Whitney U or analysis of variance for repeated measures
and subsequent Bonferroni post hoc test. P<0.05 was
considered to be statistically significant.

Statistical Analyses

Incidences and spot volumes of each protein in control and CF groups,
determined by image analysis, were compared by Fisher's exact test
and Student t test, respectively. The Student t test was based on twoway
analysis of variance, which included terms for disease status, age
(adult vs. child), and their interaction. Strong control of the familywise
type I error rate for each term (age, disease status, and interaction)
was maintained by Holm's adjustment (18), applied to all the protein
spot volumes. This procedure is a stochastically dominant modification
of the Bonferroni procedure for testing each hypothesis at a p value
of 0.05/K, where K is the number of protein spots analyzed. Statistical
analyses were performed with S-Plus 6 software (Insightful Corporation,
Seattle, WA).

Differences between biomarker concentration and disease groups.
Data were analyzed with a linear model, applied separately for children
and for adults. The linear model for children was a fixed effects linear
model, with a term representing disease status. The mixed model was
fitted with the residual maximum likelihood algorithm (19) as implemented
in the R package nlme (20) (see the online supplement for
further detail).

Statistics

Data are expressed as meansąSE. One-way ANOVA
for repeated measures was used to test for differences. The posttest
was computed only if the overall P was < 0.05. We relied on the
Bonferroni post hoc test. Differences between preflight and in-flight
values after 5 mo in space were compared by the nonparametric
Wilcoxon matched-pairs test. Linear regression analysis was used if
indicated. A value for P<0.05 was considered significant.

Statistical analysis

Microparticle levels were expressed as
median and range and analyzed using nonparametric
Mann-Whitney U tests. Thrombin-generating activity
(TGA) data were expressed as meanąSEM, according to
the distribution normality, and analyzed by analysis of
variance followed by Bonferonni post-test (StatView4.51,
SAS Institute, Cary, North Carolina). Differences were
significant at p<0.05.

Statistical Analysis

Data are expressed as the meanąSEM. Differences in clinical and
biochemical parameters between lean and obese women were determined
using the Wilcoxon unpaired nonparametric test. Spearman
coefficients were computed to examine correlations. The cellular
experiments were performed at least 3 times. Statistical analysis was
performed using Student t test. Comparisons between >2 groups
were performed using 1-way ANOVA analysis followed by post hoc
test, in which P<0.05 was considered statistically significant.