I do not know what you mean by SPSS (likely a statistical software package) but a comparison of river water quality at 2 sites needs a large set of data representative of different climatic and hydraulic conditions. Furthermore the selected quality parameters needs to be defined according to the scientific question that you want to address and to the river basin: major organic and inorganic components, metals, PAH, PCBs, endocrine disruptors, drug residues...

To start with, you need to have specifc water chemistry variables e.g. DO, Mn, Ni, etc and then compare these variables (upstream versus downstream). Makesure you have enough data points. Test your data for normality. Use parametric start e.g. ANOVA, etc if data are normally distributed or non-parametric test eg.Kruskal -Wallis test etc test to examine statistical differences if data do not meet assumption of normality. So indicate which test you wish to useand how to do it using SPSS will be simple.

First of all perhaps your question is not clear. Anyway if you meant about the concentration of different water quality parameters like DO, pH, BOD, COD, temperature and chemical parameters or heavy , metal and if the concentration as you mentioned is significantly different; which is normal in most of the cases. You can do Principal Component Analysis (PCA) using SPSS. PCA is the tool that can analyse the data having more fluctuation.

First of all this question on upstream and downstream concentrations of DO, BOD, COD,pH and certain metals variations are the flume dispersions are normally modelled in Environmental Engineering by using Streeter Phelps equations of conservative and non conservative models that includes Oxygen Sag curve analysis also.If you refer the following books there are examples of river and stream water quality modelling of pollutants in upstream and down stream areas. Principles of Environmental Engineering by Gilbert A Masters and or Princiles of Environmental Engineering by O'Corner A Dobbin book has lots of river water quality modelling examples before you proceed on statistical approach.

Dr.P.Shanmugam, Principal Scientist, Environmental Engineering, CLRI.(CSIR) Adyar, Chennai.20, Former PhD student from Leeds University.

Thank you very much for all your contributions.

To clarify on the question, I measured the hydrochemistry (EC, pH, NO3, NH4, cations, anions etc) of a river upstream of an informal settlement and downstream of the informal settlement. I want to know if this informal settlement significantly affects the water quality downstream. As a start, one can use statistics to determine if the concentrations upstream are significantly different (p

Analysis of Variance can help you

Test for normality, then use simple ANOVA if data are normally distributed or use non-parametric stats if data do not meet normality. Again, you cannot use chi-test for data of this nature.

thanks Nelson and Ashwani.

Hey Gregory thanks for these links. Initially, I was also thinking of a t-test. But there is a condition to use a t-test. The two groups of data have to be independent. Is this the case of this type of data?

I am going to read through the links you have sent

I am biologist and in my point of view, I suggest involving other parameters measure like a biochemical oxygen demand (BOD) and biological indicators. I mean, species used to monitor the health of an environment or ecosystem. They are any biological species or group of species whose function, population, or status can be used to determine ecosystem or environmental integrity, then put them on your statistical analysis according to your statistical method applied.

T test can be used for both dependent and Independent variables. The problem with t-test is that you cannot compare more than two things at the same time (although in this case you only comparing two i.e. up and down streams, I would rather still advise you use ANOVA

Nelson: I think a T-test is specifically for independent data. But most probably, a simple ANOVA is sufficient for this problem

Mr. Nyenje, If data on concentrations in the upstram of a river are collected for a certain period and for the same period the data might be collected on the concentrations in the downstream of a river. If I am right, The number observations on the upstream concentrations and downstream concentrations should be same. Therefore you can apply for a paired sample t-test where 'n' is same for two categories. The test is available under Analysis - Compare means in SPSS.

Dear Padma, yes you are right about the data. It was collected for the same periods.

Can I also use the t-test to compare means? If not, why?

Yes you can use t-test. There are different types of t-tests for comparing two-means. for your data the paired t-test is suitable.

As pointed out by earlier contributors, You can use t-test to ascertain the significance level of concentrations between the upstream and downstream,thus between sites, you could also check for seasonality-dry and wet season

thanks Emmanuel

Hi Philip Nyenje

I am agree with Padma to use the paired t-test for your case. I think it is very common being difference between the upstream and downstream concentration. Recognition of parameters cause the differences is important in such cases that I am sure you did think about it.

Regards,

Rasoul

The other problem may be that the differences are due to natural processes. So need to be careful with the results after analysis.

Regards

MAybe you can use factor analysis for the determining differences.

Statistical Methods in Water Resources by D.R Helsel & R.M. Hirsch would be useful