統計分析

採樣所得之微生物濃度以及環境因子數據係利用 Microsoft Excel 進行 建檔與管理，再使用 Microsoft Excel 或是 SPSS (SPSS Inc., USA) 進行統計 分析。由於統計方法十分多樣化，本研究根據其他文獻採用的方法 (整理於 表 3-3) 、以及根據數據資料性質與呈現方式，選擇下述幾種統計方法進行 分析：

1. 先利用 Shapiro-Wilk statistical test 檢測真菌與細菌濃度是否成常態分布，

由於檢測結果非呈現常態分布，故需使用無母數統計法進行濃度差異比 較與影響因子分析，其適用於小樣品數的資料分析，而無母數檢定方法，

又可根據資料種類及樣本之間的相關性，分類成如下表 3-2，其中次序 資料係指資料具等級含意，例如有優、佳、差、劣之分，名義資料則指 資料無等級之分，如男性與女性 (榮，2009)，而下述統計方法的結果均 採用 95%信賴區間的雙尾檢定。

2. 對於同醫院不同採樣場所、不同醫院同類型的採樣場所微生物濃度，使 用敘述統計以及 Mann-Whitney U test 的方式進行濃度差異比較與說 明。

3. 利用 Kruskal-Wallis test 分析採樣場所的室內真菌與細菌濃度於不同的 月份，是否有顯著的濃度差異。

4. 使用 Spearman correlation coefficient 分析微生物濃度與溫濕度、室內人 口數量、CO₂濃度、室外微生物濃度的相關性，評估採樣場所可能造成 IAQ 劣化的原因。

5. 採用 Wilcoxon signed rank test 比較微生物濃度於相同採樣地點，在早、

午、晚不同時段，是否有顯著的濃度差異。

表 3-2 無母數檢定分類

單一樣本 獨立樣本 相關樣本

樣本分

類組數 二個 K 個 二個 K 個

次序

資料 Kolmogorov-Smirnov Mann-Whitney Kruskal-Wallis Wilcoxon Friedman 名義

資料 Chi-Square Chi-Square Mc Nemar Cochran Q

資料來源：榮，2009

表 3-3 文獻常用統計方法

Location Variables Statistical technique reference

underground railway stations

evaluate the normality of bacteria concentrations Shapiroe-Wilk test (normality test)

Hwang et al., 2010 evaluate the differences in bacteria concentrations due to each of six

factors that were evaluated

nonparametric chi-square test

compare bacteria concentrations for the other related factors Pearson correlation

many public buildings

evaluate the differences in bioaerosol concentrations between the different buildings

ANOVA

Kim and Kim, 2006 Duncan’s test

evaluate the differences in bioaerosol concentrations between indoor and outdoor air

Means of paired t-test

six single-family

homes

evaluate the differences in fungi concentrations between indoor and outdoor air

Wilcoxon signed rank test

Lee et al., 2006 compare fungi concentrations for meteorological factors Spearman Correlation

coefficients evaluate the normality of culturable fungi or total fugal spores Shapiro-Wilk test

hospital the relationships among total particle number, and bacterial and fungal concentrations

Spearman Correlation

coefficients Li and Hou, 2003

office buildings

evaluate the normality of the data Kolmogorov–Smirnov test

Salonen et al., 2007 evaluate the differences in bioaerosol concentrations between

mould-damaged and control buildings Mann-Whitney U test

表 3-3 文獻常用統計方法 (續)

Location Variables Statistical technique reference

passenger cars and public

buses

evaluate the normality of the data Shapiro–Wilk

statistical test

Lee and Jo, 2005 characterize the lognormally distributed data Geometric Standard

Deviation

compare the temperature and humidity data sets ANOVA

compare bioaerosol data for vehicle type and seasons Wilcoxon signed rank test evaluate the differences in bioaerosol concentrations between indoor

and outdoor air t-test

many public buildings

evaluate the differences between yearly, seasonal, and regional fungal

concentrations for both indoor air and outdoor air Kruskal-Wallis test Shelton et al., 2002

Outdoor

evaluate the differences in bioaerosol concentrations between the two

sampling locations Mann-Whitney U test

Wu et al., 2007 compare ambient fungi for environmental parameters Multiple regressions

hospitals

evaluate the differences in bioaerosol concentrations between the

different hospitals Kruskal-Wallis test

于，2007 evaluate the differences in bioaerosol concentrations between the

related factors

Mann-Whitney U test Wilcoxon signed

rank test