# Need help on one-way ANOVA (SPSS) for Water Quality Analysis

#### RienaShin

##### New Member
Hi all,
I am new to statistics here. Currently, I am assessing the effect of aeration system on water quality in aquaculture ponds. I had 3 treatments (a,b,c) with 11 water quality parameters. However, these water quality parameters are measured on daily, every 5 day and weekly basis.

Daily: DO, pH, temperature
Every 5 day: Alkalinity, TAN, NO2-N, NO3-N, PO4-P
Weekly: Turbidity, SS, TSS

I am using SPSS one-way ANOVA analysis to analyze these parameters.

1) Should I separate these parameters in separate sav.files based on the measurement frequency (Daily, Every 5 day and Weekly) or should entry all these parameters in one sav.file? If it's in one sav.file, there will be missing data for the parameters on every 5 days and weekly basis, right? How to solve the problem?

2) In SPSS, under one way ANOVA, should I compute one dependent variable (eg. DO only) at a time for 3 independent groups or choose all dependent variables (eg, DO, pH, temp) for 3 independent group?

3) How to identify ouliers before ANOVA analysis?

4) Which test is recommended to check whether dependent variable (water quality) is normally distributed for independent variables? (skewness & kurtosis, Sharpiro-wilk, histogram, Normal Q-Q)

5) What to do if the homogeneity of variance is violated (p=0.000)? I have checked for Welsh and Brown-Forsythe, but similarly both tests also yield p=0.000. Does that mean there is no equal variance assumed? There is a statement stated "as long as N>30 and n1 is approximate to n2, test is robust to violation of homogeneity of variance"? Should I proceed with one-way ANOVA analysis or change to Kruskal Wallis method?

Please forgive me if any of my words are confusing, as I am very new to statistic fields. I would be grateful if any advise and help is given.
Thank you very much.

#### abhay9318

##### New Member
3)You can plot box -plot in SPSS to find the outliers.